Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Retrospective Theses and Dissertations Iowa State University Capstones, Theses andDissertations
1998
An examination of the effects of personality and jobsatisfaction on multiple non-workroleorganizational behaviorsDouglas Dale MolitorIowa State University
Follow this and additional works at: https://lib.dr.iastate.edu/rtd
Part of the Industrial and Organizational Psychology Commons, and the Personality and SocialContexts Commons
This Dissertation is brought to you for free and open access by the Iowa State University Capstones, Theses and Dissertations at Iowa State UniversityDigital Repository. It has been accepted for inclusion in Retrospective Theses and Dissertations by an authorized administrator of Iowa State UniversityDigital Repository. For more information, please contact [email protected].
Recommended CitationMolitor, Douglas Dale, "An examination of the effects of personality and job satisfaction on multiple non-workrole organizationalbehaviors " (1998). Retrospective Theses and Dissertations. 11635.https://lib.dr.iastate.edu/rtd/11635
INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter fece, while others may be
from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper aligrmient can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in reduced
form at the back of the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6" x 9" black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly to
order.
UMI A Bell & Howell Infonnation Company
300 Noith Zetb Road, Ann Arbor MI 48106-1346 USA 313/761-4700 800/521-0600
An examination of the effects of personality and job satisfaction on multiple
non-workrole organizational behaviors
by
Douglas Dale Molitor
A dissertation submitted to the graduate faculty
in partial fulfillment of the requirements for the degree of
DOCTOR OF PHILOSOPHY
Major: Psychology
Major Professor: Kathy A. Hanisch
Iowa State University
Ames, Iowa
1998
Copyright © Douglas Dale Molitor, 1998. All rights reserved.
TJMI Number; 982 6558
UMI Microform 9826558 Copyright 1998, by UMI Company. Ail rights reserved.
This microform edition is protected against unauthorized copying under Title 17, United States Code.
UMI 300 North Zeeb Road Ann Arbor, IVn 48103
ii
Graduate College Iowa State University
This is to certify that the Doctoral dissertation of
Douglas Dale Molitor
has met the dissertation requirements of Iowa State University
•Major Professor
For the Major Program
For the Graduate College
Signature was redacted for privacy.
Signature was redacted for privacy.
Signature was redacted for privacy.
iii
TABLE OF CONTENTS
ABSTRACT iv
CHAPTER I: INTRODUCTION I
CHAPTER 2: LITERATURE REVIEW OF ATTITUDES AND BEHAVIORS AT WORK 4
CHAPTER 3: LITERATURE REVIEW OF PERSONALITY AND BEHAVIORS AT WORK 24
CHAPTER 4: METHOD 54
CHAPTERS: RESULTS 70
CHAPTER 6: DISCUSSION 91
REFERENCES CITED 134
iv
ABSTRACT
This study examined the multiple relationships between job satisfaction, personality,
and non-workrole behaviors. Non-work role behaviors are defined here as groups of positive
and negative behaviors that influence organizational effectiveness but are not part of a formal
job description or controlled by an organization's reward/performance evaluation system. In
recent years, both job satisfaction and personality have received renewed research attention
examining how they contribute to the explanation and prediction of traditional organizational
criteria such as job performance and training success. This study used the five-factor model
of personality (Digman, 1990), job satisfaction, and positive and negative affect to explain
employees' non-workrole behaviors. Using a self-report siuvey. data were collected firom
313 employees in the health care industry. Two stage structural equation modeling was used
to compare different theoretical models evaluating the contribution of job satisfaction.
positive and negative affect, and alternate conceptualizations of personality to the prediction
of non-workrole behaviors. The results suggest that job satisfaction, affective state, and
personality contribute uniquely to the prediction of non-workrole behaviors. The results also
suggest that criterion-related conceptualizations of personality are more successful in the
prediction of non-workrole behaviors than more general conceptualizations of personality. In
addition to these findings, support is also provided for the congruent measurement of general
attimdes and general behaviors (i.e., behavioral families). Theoretical and practical
implications are discussed.
I
CHAPTER 1: INTRODUCTION
Industrial and organizational (I/O) psychology is largely occupied by the task of
determining how differences between employees' attributes affect behavior at work. The
complexity of human behavior makes this an extremely difficult task. Employees can differ
on many dimensions; each dimension having different effects on their behavior. Common
dimensions where differences exist include intelligence, work attitudes, and personality and
they all have received significant research attention. These areas of research have, however,
produced decidedly different results.
The research on intelligence and cognitive ability has been highly successful. Few-
would argue with the conclusion that cognitive ability is probably the single best variable
available if one is attempting to predict an applicant's future performance behavior. For
example, in a large study involving military personnel, validity coefficients for general
cognitive ability predicting general soldiering and technical proficiency were reported to be
approximately .65 (McHenry, Hough, Toquam, Hanson, & Ashworth, 1990). Cognitive
ability has also been shown to be a valid predictor of employee performance across nearly all
jobs and situations (Hunter. 1986; Schmidt & Hunter, 1981).
Research on the usefulness of attitudes and personality in predicting relevant
employee behavior has not produced the same level of success as cognitive ability. Although
job satisfaction is probably the most commonly investigated work-related attitude and has
been researched extensively (see Cranny, Smith, & Stone, 1992), it has produced little
empirical evidence showing a relation to employee performance (Vroom, 1964; laffaldano &
Muchinsky, 1985; Podsakoff & Williams, 1986). Research on the relation between job
satisfaction and individual withdrawal behaviors such as absenteeism or turnover, has
produced slightly better, yet still moderate results (Fisher & Locke, 1992). Although job
satisfaction remains a popular topic of research, there are those who have questioned the
usefulness of the job satisfaction construct (Salancik & Pfeffer, 1977; 1978).
-7
Like job satisfaction, the criterion-related conclusions offered from personality
research have also been less than impressive. In fact, they have generally been described as
disappointing (Ghiselli & BarthoL 1953: Guion & Cottier. 1965). A lack of predictive
success and disagreement among researchers regarding the definition and conceptualization
of personality mainly led to its abandonment as a plausible explanatory variable in I/O
psychology.
Recentiy, however, job satisfaction (Cranny et al., 1992; Hulin. 1991) and personality
(Hogan. 1991; Schneider & Hough, 1995) have once again begun to receive attention from
I/O psychologists. Conceptual and methodological improvements in job satisfaction research
have led to new hope in understanding its usefulness (Guion, 1992a). In fact. Roznowski and
Hulin (1992) suggest that job satisfaction is likely the most important variable in predicting
employee behavior c^er individuals have been hired (cognitive ability would be considered
the best predictor/jrzor to organizational entry). The changes in job satisfaction research
have focused on the conceptualization and measurement of multiple behavioral criteria. In
the past, most job satisfaction research attempted to use employees' overall attimde toward a
job to predict one or two specific behaviors such as output or absenteeism. Looking at such a
narrow band of the wide spectrum of behavioral options available to employees, has been
argued to have resulted in the relatively poor behavioral prediction using overall attitudes
(Hanisch. 1995a; Hulin, 1991).
Personality research has also experienced somewhat of a revival in organizational
research (KatzelL 1994). Much of the renewed interest comes from the popularity of the
five-factor model of personality (also referred to as the "Big Five") that represents the most
basic dimensions of personality. Studies examining personality in organizational settings
have demonstrated that elements of the five-factor model provide incremental validity in the
prediction of employee performance beyond what is currentiy possible using tests of
knowledge, skills, or abilities alone (Schneider & Hough, 1995; Hogan. 1991; Barrick &
J
Mount, 1991; Rosse, Miller, & Barnes. 1991: TetL Jackson. & Rothstein. 1991; Barrick &
Mount, 1993).
This study examined the contributions of job satisfaction, affect, and personality to
the explanation of the general behavioral constructs of organizational withdrawal and
organizational citizenship. Briefly, organizational withdrawal refers to behaviors employees
engage in to remove themselves from their workroles or the job itself (Hanisch. 1995a).
Organizational citizenship behaviors (OCB) refer to positive behaviors employees engage in
at work that are not part of their job but contribute to the organization's effectiveness.
Together, these two types of behavior can be thought to compose a single, larger group of
positive and negative behaviors to be referred to here as non-workrole behaviors. Non-
workrole behaviors then are defined as those employee behaviors that are not part of a job
description and are not necessarily governed by an organization's reward system or
performance evaluations. Examples of these types of behaviors include making frequent
trips to the water fountain or restroom. gossiping with co-workers, leaving work early,
helping new employees, or making suggestions on ways to improve things at work.
The following literature review provides information regarding research relevant to
job satisfaction, affect, personality, and non-workrole behaviors as they are used in this
study. First, theoretical and conceptual issues relating to the measurement of attitudes and
behaviors at work are addressed. This section includes job satisfaction and its relation to
organizational withdrawal and organizational citizenship. The next section will introduce
personality and a brief history of its use in I/O psychology. The discussion of personality
also involves the consideration of positive and negative affect as constructs that are similar,
yet distinct from personality. Past research documenting the relations between personality,
affect, job satisfaction, and employees' non-workrole behaviors are also described. Finally,
the section concludes with hypotheses about expected results as well as the theoretical and
practical contributions of the current study.
4
CHAPTER 2: LITERATURE REVIEW OF ATTITUDES AND BEHAVIORS
AT WORK
Given the long history of job satisfaction research, it is difficult to bring something
novel to its study. Job satisfaction is certainly one of the most thoroughly researched topics
in I/O psychology. In 1976, Locke estimated that over 3,000 articles had been written on the
subject and researchers have continued to produce studies over the last 20 years.
Researchers have examined its relation to a plethora of organizational variables including, for
example, motivation, productivity, commitment, absenteeism, and quitting (see Locke, 1976
for a review). Because of its assumed importance, researchers have vigorously studied job
satisfaction in an attempt to determine both its causes and its consequences. This research
has produced many different theories of job satisfaction. There is, however, disagreement
among researchers regarding which theory is best.
Despite the lack of agreement about which particular theory is best, the most common
theories of job satisfaction are based on the employee's expectations or values and how well
those expectations or values are being fulfilled. This provides for some general agreement
among job satisfaction researchers regarding what the construct of job satisfaction. Cranny et
al. (1992) state that job satisfaction "is an affective (that is, emotional) reaction to a job that
results from the incumbent's comparison of actual outcomes with those that are desired
(expected, deserved, and so on)" (p. 1). In other words, job satisfaction is an emotional state
(or attitude) that results from an employee's evaluation of the degree to which his or her
expectations are being met.
Job satisfaction has generally been conceptualized as a single general construct. Job
satisfaction research has. however, frequently addressed how job satisfaction is related to
multiple facets of the job. Smith. Kendall, and Hulin (1969). for example, have demonstrated
that job satisfaction can be concepmalized as a general construct that is composed of five
distinct facets. These facets are satisfaction with work, satisfaction with pay and benefits.
5
satisfaction with co-workers, satisfaction with supervision, and satisfaction with promotion.
Weiss. Dawis, England, and Lofquist (1967) also have conceptualized job satisfaction as a
multi-faceted construct. Their job satisfaction scale measures 20 facets of work including
working conditions, independence, technical-supervision, human relations-supervision, and
others.
Beyond the question of what causes job satisfaction, the most frequently asked
questions concern the consequences of job satisfaction and how behaviors in organizations
are affected by dissatisfied employees. To answer these and other questions, thousands of
studies have been conducted examining the correlates and consequences of job satisfaction.
The most popular behavioral criteria used in job satisfaction research are employee
performance and attendance/turnover variables (Locke, 1976).
Establishing the relationship between job satisfaction and performance has been
especially troublesome for researchers throughout the history of its study. Although it is a
commonly held belief among people in general that employees who are satisfied with their
jobs will perform better than employees who are not satisfied, it has yet to be adequately
demonstrated through empirical research. Brayfield and Crockett (1955) have been credited
with calling researchers' attention to the weak associations between job satisfaction and
performance (Katzell, Thompson, & Guzzo, 1992). Twenty-one years later. Locke (1976)
described the literature as consistently providing "negligible" relationships between job
satisfaction and productivity. More recent investigations have found similar results. For
example, a meta-analytic review by laffaldano and Muchinsky (1985) produced an average
correlation between job satisfaction and performance of .14. In another review, Podsakoff
and Williams (1986) reported a similar correlation between job satisfaction and performance
of .17. These correlations suggest that it is quite obvious that the relationship between job
satisfaction and performance is not as simple or direct as is commonly believed.
6
Although work performance has not proven to be a good criterion for establishing the
effects of job satisfaction, there has been somewhat greater success in demonstrating the
negative relationships between job satisfaction and employee behaviors such as absenteeism
and tumover. Even though the correlations are still not very large, job satisfaction has been
shown to affect employees' decisions regarding behavior such as attendance, leaving an
organization, and retiring (Hackett, 1989; Hanisch & Hulin, 1990; Tett & Meyer. 1993).
Hackett reported correlations between general job satisfaction and absence to be approximately
-.20 (r = -. 15 with frequency of absence and r = -.23 for duration of absence). Tett and Meyer
reported a correlation of -.25 between general job satisfaction and tumover. Hanisch and
Hulin reported correlations between work satisfaction and desire to retire to be approximately
-24.
While these correlations are higher than those reported between job satisfaction and
performance, they are still moderate at best. The problem of a low attimde-behavior
correlation is not unique to job satisfaction research. Social psychologists have been aware of
the "attitude-behavior" problem longer than I/O psychologists and have exerted a significant
research effort in an attempt to understand it.
Attitudes and Behaviors
The attitude-behavior problem has been around almost as long as psychologists have
been measuring attitudes (Petty & Cocioppo, 1981). As early as the 1930's, social
psychologists had noted attimde-behavior inconsistencies between participants' responses on
attimde surveys and their actual recorded behaviors (e.g., Corey. 1937; LaPiere. 1934). The
lack of expected correlation between attitudes and behaviors continued and eventually caused
researchers to question the value of the entire attimde concept. Wicker (1969) reviewed over
30 articles on the attitude-behavior relationship and concluded that "taken as a whole, these
smdies suggest that it is considerably more likely that attitudes will be unrelated or only
slightly related to overt behaviors than that attitudes will be closely related to actions"
7
(p. 65). Based on this lack of successful behavior prediction. Wicker eventually called to
"abandon the attitude concept" (1971. p. 29). Fortunately, the attitude concept was not
abandoned, and researchers instead continued working to better understand the relationships
between attitudes and behaviors. Since Wicker's call to abandon the attitude concept, some
of the most influential work in the area of attitudes and behaviors was conducted by Fishbein
and Ajzen (1974; 1975) which resulted in the theory of reasoned action and its later
incarnation—the theory of plarmed behavior (Ajzen. 1991).
Briefly, both the theory of reasoned action and the theory of planned behavior suggest
that behavior is determined by an individual's intention to perform a particular behavior.
They both state that the intention to perform a behavior is determined by a combination of
the individual's attitude toward performing a behavior and his or her subjective social norms
regarding the particular behavior. The distinctions between the two theories are not
important to this study. The following discussion includes the elements that are common to
both theories and are important to understanding the attitude-behavior relationship.
One of the most important tenets and necessary conditions for both the theory of
reasoned action and the theory of planned behavior is the issue of correspondence between
attitudes and behaviors. That is. the level of specificity or complexity at which attitudes and
behaviors are defined and measured must be equal. This is what Ajzen and Fishbein (1977)
refer to as the correspondence between attitudinal and behavioral entities. Much of the
existing attitude research fails to match the level of specificity between the attitude and the
behavior to be predicted. This incongruence makes prediction of behaviors difficult and
results in low correlations (Hulin, 1991). For example, Weigel, Vernon, and Tognacci (1974)
measured individual's general attitudes regarding the importance of a clean environment and
then recorded their behaviors when given the opportunity to participate in specific volunteer
activities with the Sierra Club. The correlation between participants' general attitude toward
the environment and specific volunteer behaviors such as writing a letter to an elected official
s
or serving on a Club committee was, as tiie authors expected, quite small (/• = .06). As a
comparison, Weigel et al. also used a high specificity measure of the participants' attimdes
regarding specific attitude objects such as the Sierra Club and participating in its activities.
The use of these corresponding measures resulted in a much stronger relationship between
the specific attitudes and behaviors (r = .68). These results illustrate the importance of
having congruent measures.
Davidson and Jaccard (1979) also conducted a study demonstrating that stronger
relationships than those reported previously result fi-om the corresponding measurement of
attitudes and behaviors. Their study measured women's attitudes, in varying degrees of
specificity, toward the use of oral contraception. For example, at the general attitude level,
participants were asked about their attitude toward birth control. At the specific level,
participants were asked about their attitudes toward using birth control pills during the next
two years. These attitudes were then correlated with reported contraceptive use. The
reported correlation between oral contraceptive use and the general attitude toward birth
control was .08. whereas it was .57 between oral contraceptive use and the specific attitude
toward using birth control pills over the next two years.
Evidence firom the Weigel et al. (1974) and Davidson and Jaccard (1979) studies
demonstrate the importance of congruence between measures of attimdes and behaviors. All
of the evidence, however, focused on matching specific measures of attitudes with specific
measures of behaviors. There was no attention being given to the other side of the
correspondence issue. That is, correspondence between measures of general attitudes and
general behaviors. This is a basic idea firom attimde theory that was originally offered by
Thurstone (1931) and reiterated by Doob (1947) several years later. By focusing exclusively
on specific attitudes and specific behaviors, the tenet of correspondence was not being fully
utilized. Much more could be gained by taking the concept of correspondence the opposite
direction and matching attitudes and behaviors at a general level zis opposed to a specific
9
level. Using specific attitudes and specific behaviors is sufficient when the goal is simple,
empirical prediction. However, the prediction of specific behaviors from specific attitudes
does not necessarily help in the understanding of behaviors. Hanisch (1995a) and Hulin
(1991) suggest that using general constructs and general behaviors will help provide an
understanding of the behaviors beyond that achieved by examining specific constructs and
specific behaviors. That is, to better understand behavior, it is beneficial to look beyond
specific instances and examine the broader theoretical underpinnings of groups of behaviors.
General Behavioral Classes
The idea of focusing on general attitudes and general behaviors has existed for some
time. In an early article on the measurement of attitudes, Thurstone (1931) stated that it is
not the specific behaviors that should be of concern as much as the characteristics or features
of the behaviors (i.e., the meaning or fiinction of the behaviors). He points out that two
people can have attitudes that are equally favorable toward an object, yet their overt action
toward the object can take quite different forms. In other words, more attention should be
given to the overall favorableness or unfavorableness of the behaviors toward a given object
instead of one specific form of behavior.
Doob (1947) also argued diat single, specific behaviors are seldom predicted from the
sole knowledge of an individual's general attitude toward an object. Conceptually, this
means that researchers need to focus on a broad spectrum of behaviors that are expected to be
related. Methodologically, this means creating behavioral measures that assess groups of
behaviors that have a similar function or meaning to the individual. These general groups
(i.e., families or classes) of behaviors should be the intended targets for prediction using
measures of general attitudes. While predicting specific behaviors (e.g., predicting who will
be absent and when) is useful to an organization, given the complexity of human behavior,
predicting specific behaviors from general attitudes is an unrealistic goal. It should, however,
be more likely that a behavioral family or class is predictable even if its specific
10
manifestation is not. Lubinski and Thompson (1986) discuss the basic units of human
behavior and communicate a similar message about studying a broader level of behavior than
is typically examined. They suggest that behavioral classes composed of more specific
fundamental units should be the behavioral constructs of interest throughout psychology.
They state that aggregates of individual behaviors are entities in their own right and can be
used as units of analysis. Although the individual behaviors composing a behavioral family
may take different forms, they are serving similar fimctions and likely have similar
antecedents.
Lubinski and Thompson illustrate their pomt by comparing the logic of behavioral
aggregation in psychology with an example from chemistry. They state the following:
Focusing our attention on prediction response classes (as opposed to the constituent
components of response classes) is consistent with the manner in which other sciences
operate. A chemist, for example, is able to predict with great certainty the reaction
produced by mixing various solutions, provided their respective volumes are known
beforehand. If. however, a specific molecule in a solution was radioactively labeled
before it was mixed, and if a chemist were asked if this particular (molecular) entity
will be involved in the (molar) reaction, our chemist's reply would be much less
precise (i.e., the molar phenomena is quite predictable but the specific molecular
constiments are indeterminate). (1986. pp. 308-309)
The above analogy relates to the smdy of work attitudes and employee behaviors
because it illustrates how organizational as well as other researchers are too often only
concerned with a single molecular behavior such as absenteeism or quitting. They fail to
consider the entire "solution" of related behaviors that may be serving a similar function
representing an underlying trait or attitude for the individual.
The principle of aggregation has also been clearly demonstrated by Rushton.
Brainerd. and Pressly (1983). Rushton et al., using examples of studies from several
II
subdisciplines of psychology, demonstrated how relationships can be established by
aggregating the variables involved. They suggested that previous studies had not
successfully demonstrated relationships because of inadequate, single-variable measurement
of the constructs involved. They point out that the sum of several measurements of a given
topic will be a more stable and accurate representation of an underlying construct than any
single measure. This occurs because combining several measures reduces the effects of
measurement error. This results in random error effects being averaged out. while the effects
of accurate measurement continue to accumulate.
Using families of aggregated behaviors as criteria is consistent with Ajzen's theory of
planned behavior as well as others that preceded it (e.g., Doob, 1947; Fishbein & Ajzen.
1974; Thurstone. 1931). The principle of aggregation is certainly not new. and its usefulness
has been demonstrated in several areas such as developmental, personality, and social
psychology (See Rushton et al.. 1983). It simply has not been utilized to the extent that it
should be. Behavior prediction in industrial and organizational psychology has for too long
focused on individual, isolated behavior. The idea of aggregating responses and considering
broad behavioral families should more frequently be applied to studies of behavior at work to
be sure that the entire spectrum of employee responses are being considered. By doing so.
researchers can better understand the influences of general work attitudes such as job
satisfaction on behavior at work.
Job Satisfaction and Behaviors at Work
As previously discussed, job satisfaction has been described as less than impressive in
its ability to predict employees' single behaviors. A likely reason for its poor performance
has been an inappropriate focus on specific behaviors such as output, performance,
absenteeism, or turnover given its status as a general attitude.
Hulin (1991) and others (e.g.. Fisher & Locke, 1992; Hanisch, 1995a; Hanisch &
Hulin, 1990; 1991; Hanisch, Hulin, & Roznowski. in press, Roznowski & Hulin 1992) have
12
made strong arguments regarding how job satisfaction research needs to be changed to better
demonstrate how it affects employee behavior. The issue is related to the congruence
between the level of specificity and complexity at which both job satisfaction and behaviors
are measured. For example, job satisfaction-withdrawal studies in the past have tended to
focus on how an employee's general feelings about his or her job is related to specific isolated
behaviors such as absenteeism or turnover. As was discussed above, trying to predict
specific behaviors from general attitudes is inappropriate and therefore, the results are not
useful given the incongruent assessment of attitudes and behaviors.
The following analogy helps to clarify the necessity of using corresponding levels of
attitude and behavior measurement. Suppose scientists discovered life on another planet and
an effort was initiated to try to communicate with them. Based on existing extraterrestrial
communication theories, it was assumed that the new life forms were aware of general radio
technology. Therefore, the scientists on earth sent messages to the new life forms using all
available forms of radio waves: AM. FM, short-wave, microwave, etc. However, when the
time came to listen for a reply, the scientists only monitored two frequencies on only one
type of radio receiver. When no response was received on these two monitored channels, it
was assimied that the new life forms did not have the ability to communicate via radio waves.
This finding caused the scientists to abandon the theories that led them to believe that the
aliens were aware of radio technology.
The scientists' levels of specificity in sending and receiving radio transmissions were
incongruent. When sending their message, they used general radio technology. When they
were listening for a response, they used specific radio technology. This is similar to
psychologists who assess employees' general attitudes about aspects of their jobs, but only
record one or two of the many possible specific behavioral responses.
The present study incorporated the concept of broad behavioral classes by examining
a wide variety of behaviors and their relation to different levels of job satisfaction or
13
dissatisfaction. Of particular interest to this study are the positive and negative behaviors that
are not normally associated with traditional measures of performance (i.e.. non-workrole
behaviors) in organizations.
Non-Workrole Behavior
Non-workrole behavior is a term that is used here to refer to employee behaviors that
have generally been studied separately under different names such as prosocial organizational
behavior (Brief & Motowidlo, 1986), OCB (Smith, Organ, & Near, 1983), organizational
withdrawal (Hanisch, 1995b; Hanisch & Hulin, 1990), and noncompliant behaviors (Puffer.
1987). These concepts are similar because they focus on behaviors that are not part of
employees' traditional workroles. Therefore, they are each included as criteria in the study
presented here.
Withdrawal Research. To an even greater degree than other areas of psychological
research, the history of withdrawal research shows that most investigators have focused
mainly on a few high-profile behaviors such as absenteeism and turnover. Absenteeism and
turnover are of importance to organizational managers because of the extreme costs
associated with them. Steers and Rhodes (1978) claimed that absenteeism costs
organizations billions of dollars each year. Turnover can causes similar costs because of the
e.xpenses associated with continually recruiting, selecting, and training new employees
(Dipboye, Smith, & Howell, 1994). Because of this financial importance, the question of
when, how. and why people withdraw from their jobs has been the subject of much research
and has produced several theoretical models. For example. Steers and Rhodes (1978),
Mobley. Homer, and Hollingsworth (1978), March and Simon (1958), and Price (1977) all
offer models of turnover or absenteeism. These examples do not exhaust the list of proposed
models, but they do represent those theories that have received most of the attention in the
withdrawal literature.
14
Although each of the models is different most of these models take a similar
approach to mapping out the particular influences on employees' decision making processes
regarding absenteeism and turnover. The common theme among them is that each includes
some type of evaluation of the employees' inputs versus outputs to arrive at a decision
regarding being absent or not continuing membership in the organization. Hulin. Roznowski,
and Hachiya (1985) combined many of the ideas from previous withdrawal models and
provided a comprehensive theoretical model that addresses both attitude formation and
withdrawal behavior. The Hulin et al. model is consistent with theories regarding attitude
formation proposed by March and Simon (1958), Smith et al. (1969), and Thibaut and Kelley
(1959). It is also in agreement with theoretical propositions regarding the relationship
between attitudes and behaviors put forward by Ajzen (1988, 1991), Fishbein and Ajzen
(1974. 1975), and Triandis (1971).
The ideas presented by Hulin et al. (1985) are similarly presented by Hulin (1991) in
a slightly modified form from the original model (See Figure 1). He details its development
and relation to other attitude-behavior theories of withdrawal that have preceded it (e.g.,
Thibaut & Kelley. 1959; March & Simon, 1958; Smith et al. 1969; Rosse & Miller. 1984;
Hulin et al.. 1985). The model suggests that job satisfaction results from employees'
evaluation of their workrole inputs and outcomes. The evaluation of workrole inputs and
outcomes is influenced by employees' frames of reference and personal utilities of direct and
opportunity costs of workrole membership. Job satisfaction will only result if the evaluation
of workrole outcomes exceeds the perceived workrole costs.
Job dissatisfaction is assumed to be an unpleasant state that prompts employees to
take action to alleviate the discrepancy between workrole inputs and outcomes. Employees
have many behaviors from which to choose to either increase workrole outcomes or decrease
workrole inputs. Hulin et al. (1985) suggest that these behaviors make up four categories.
The first, labeled as specific attempts to increase job outcomes, includes behaviors such as
Job/Work Role Satisfaction
Behavioral Job Withdrawal/A voidanct
Psychological Job Withdrawal
Specific Change
Behaviors
Specific Attempts to Increase Job Outcomes
Work Role Inputs o Skills o Time o Effort o Training o Forgone opportunities
Behavioral Intentions to Change Work Situation o Unionization activity o Transfer attempts o Demotion attempts
Utility of Direct and Opportunity Costs 0 Local unemployment 0 Specific occupational
unemployment o Available alternatives
Work Role Outcomes
o Salary/wages o Fringe benefits o Status o Working conditions o Intrinsic rewards
Frames of Reference for Evaluating Job Outcomes
o Past experience o Local economic
conditions
Behavioral Intentions to Reduce Work Role Inclusion o Absenteeism o Turnover o Retirment
Behavioral Intentions to increase Job Outcomes o Stealing o Using work time for
personal tasks o "Moonlighting" on the job
Behavioral Intentions to Reduce Job Inputs o Missing meetings 0 Long coffee breaks o Drinking/drugging before
work o Talking trivia with
co-workers
Figure 1. llulin's theoretical model of adaptive responses (llulin, 1991).
16
stealing or using work time for personal tasks. The second category is labeled psychological
withdrawal and consists of behaviors such as taking long breaks and talking trivia with co
workers. The third category is called behavioral job withdrawal/avoidance. It consists of
more drastic behaviors such as absenteeism, turnover, and retirement. Finally, the fourth
category—specific change behaviors—refers to those behaviors that are meant to change the
work situation. Examples here include unionization attempts or talking to the boss about
changing the work itself.
Hanisch (1995a) has further developed the withdrawal aspects of Hulin's (1991)
model (i.e., categories two and three from above) and has identified and defined a broad
behavioral construct labeled organizational withdrawal. Employees use organizational
withdrawal behaviors to avoid their work or remove themselves from their jobs entirely.
Organizational withdrawal is defined by Hanisch (1995b) as "a general construct
composed of a variety of acts, or surrogate intentions, that reflect both the negativity of the
precipitating job attitudes and the target of these negative job attitudes." In other words, it is
a group of behaviors that is meant to encompass the many manifestations of withdrawal that
can result from negative work attitudes such as job dissatisfaction. By doing so, it adheres to
the recommendations cited above regarding aggregation (Rushton et al., 1983) and
congruence in the measurement of both attimdes and behaviors (Doob, 1947; Fishbein &
Ajzen. 1974; Thurstone, 1931).
The general construct of organizational withdrawal has been shown empirically using
factor analysis and causal modeling to consist of two distinct components: Work withdrawal
and job withdrawal (Hanisch, 1995a; Hanisch & Hulin, 1990,1991). Work withdrawal refers
to those behaviors that dissatisfied employees engage in to avoid or minimize the time spent
on aspects of their work while maintaining organizational and workrole membership (e.g.,
being late for work, missing meetings, daydreaming). Job withdrawal refers to those efforts
17
an employee engages in to remove himself or herself from an organization entirely (e.g..
turnover, retirement).
The distinction between work and job withdrawal is important because, as cited by
Hulin (1991), Atkinson and Birch (1970) suggest that individuals will select behaviors to
maximize the utility of performing the behavior. Studies discussed by Hulin have shown that
employees discriminate among sources of dissatisfaction and choose behaviors that are likely
to help alleviate its discomfort. For example, Getman. Goldberg, and Herman (1976)
reported that satisfaction with elements of the job most likely to be changed by unionization
(i.e., pay and supervision) were strongly related to the probability of voting for unionization
(an adaptive behavior). Schriesheim (1978) reported a similar finding, stating the correlation
between pro-union voting and economic factors of the job to be -.76. The reported
correlation between union voting and non-economic factors of the job was -.38.
If employees are selecting withdrawal behaviors that they see as being most helpful in
alleviating their dissatisfaction, it becomes very important to measure a broad construct that
will represent as many of the adaptive options as possible, not just absenteeism or turnover.
This is because, as Roznowski and Hulin (1992) state, the employee may mix and match
different behaviors from a single behavioral family until the desired outcome is achieved.
The employees' choices regarding which withdrawal behaviors to enact are govemed
by two broad influences known as response valence and response threshold (Hanisch.
1995a). Influences on response valences refers to those influences governing the positive or
negative value of the behavioral responses. For example, past experiences with the
effectiveness of calling in sick to reduce the discomfort of dissatisfaction will influence the
decision of whether to call in sick again. Influences on response thresholds refers to
influences that affect the probability of performing a behavior. For example, a poor local
economy would influence an employee's decision to quit his or her job. Because of the
18
difficulty of finding a new job, tiie poor economy would increase the threshold and thereby
decrease the likelihood of choosing that behavior to reduce dissatisfaction.
Theoretically, as these threshold and valence influences change, employees will
switch between behaviors depending on which are most appropriate for their current
adaptation needs. The proposed switching among different withdrawal behaviors fiirther
necessitates the use of broad behavioral measurement. Studying only one behavior such as
absenteeism may result in null results simply because the response threshold for that behavior
may be too great in the particular organization being studied. If the organization has a strict
absence policy, the employee may choose to be late to work instead because the
consequences are less than being absent. In this example, being late would have a lower
response threshold than being absent.
It is because of this potential switching between available responses and differences
in response valences and thresholds that individual behaviors such as absenteeism or
tardiness have been and always will be difficult to predict from general attitudes. However,
predictions of a more general behavioral construct such as organizational withdrawal will be
considerably more successful. Roznowski and Hanisch (1990) demonstrated this idea in a
study examining job satisfaction and behavioral composites consisting of several adaptive
behaviors. By systematically increasing levels of generality in the measurement of
withdrawal behavior (including withdrawal intentions and withdrawal cognitions), they
reported increased correlations between measures of job satisfaction and organizational
withdrawal. For example, the correlation reported between the JDI satisfaction with work
subscale and a composite of withdrawal behaviors was -.36. The work satisfaction subscale
was also correlated with a more general composite that included withdrawal behaviors as
well as withdrawal cognitions and withdrawal intentions. The correlation with this more
general measure of withdrawal was -.53. Similar results were reported for each of the
satisfaction subscales used in the study.
19
Based on the evidence presented above, a strong case can be made for the use of
organizational withdrawal as a general behavioral construct. In this particular study, the form
of organizational withdrawal specifically defined as work withdrawal is examined. Given the
conceptualization of non-workrole behaviors in this study as behaviors that occur at work,
work withdrawal is appropriate to represent negative non-workrole behaviors because they
consist of negative behaviors that employees engage in to remove themselves from work
while maintaining membership in the organization. The research reviewed above suggests
that examining the multiple responses that compose work withdrawal instead of one or two
specific behaviors that have traditionally been the subject of withdrawal research will provide
a better understanding of employees' responses to job dissatisfaction.
Positive Non-Workrole Behaviors. Positive non-workrole behaviors include those
behaviors that are labeled in the literature as Organizational Citizenship Behaviors (OCB)
and Prosocial Organizational Behaviors (POB). Organ and Konovsky (1989) define OCBs as
"constructive or cooperative gestures that are neither mandatory in-roIe behaviors nor directly
or contractually compensated by formal reward systems" (p. 157). OCB and POB are
important because they represent employee behaviors such as helping, cooperating, and
volunteering that, accumulated over time, are presumed to be important to organizational
effectiveness.
POB has been presented in the literature as being a more general concept than OCB.
Brief and Motowidlo (1986) offered the following definition:
Prosocial organizational behavior is behavior which is (a) performed by a member of
an organization, (b) directed toward an individual, group, or organization with whom
he or she interacts while carrying out his or her organizational role, and (c) performed
with the intention of promoting the welfare of the individual, group, or organization
toward which it is directed (p. 711).
20
From the preceduag definition, it can be seen that POBs include more than just non-
workxole behaviors. Many of the behaviors included within the above definition are
legitimate forms of job performance. Brief and Motowidlo (1986) identified 13 theoretical
(i.e., not empirically determined) dimensions upon which POBs can occur (see Table 1).
Although they provide a useful taxonomy for positive behaviors at work, many of Brief and
Motowidlo's categories go beyond the non-workrole behaviors that are the focus of this
study. Alternatively. OCB provides a conceptualization of positive organizational behavior
that is more precisely related to the non-workrole behaviors to be studied here. Organ (1988)
defines OCB as the following, "individual behavior that is discretionary, not directly or
explicitly recognized by the formal reward system, and that in the aggregate promotes the
effective functioning of the organization" (p. 4). Given this definition, OCB is recognized to
be a more specific subset of the larger FOB concept.
OCB has been shown via factor analysis to be divisible into more than one factor
(Organ & Konovsky, 1989; Podsakoff, MacKenzie, Moorman, & Fetter, 1989; Smith et al..
1983). Smith et al. (1983) claimed that there are two factors including Altruism and
Generalized Compliance. Altruism consists of personal prosocial/helping behaviors.
Generalized Compliance refers to impersonal, conscientiousness behaviors. Organ and
Table 1. Brief and Motowidlo's (1986) 13 categories of prosocial organizational behaviors.
Assisting co-workers with job-related matters. Assisting co-workers with personal matters. Showing leniency in personnel decisions. Providing services or products to consumers in organizationally consistent ways. Providing services or products to consumers in organizationally inconsistent ways. Helping consumers with personal matters unrelated to organizational services or products. Complying with organizational values, policies, and regulations. Suggesting procedural, administrative, or organizational improvements. Objecting to improper directives, procedures, or policies. Putting forth extra effort on the job. Volimteering for additional assignments. Staying with the organization despite temporary hardships. Representing the organization favorably to outsiders.
21
Konovsky (1989) conducted a factor analysis of the l6-item Smith et al. (1983) measure of
OCB and determined that the Generalized Compliance factor could be further divided into
two separate factors. They claimed that one part of the Compliance factor focused on the
conscientious use of time. The second factor consisted of three items involving negative
behaviors that the employee does not do.
Podsakoff et al. (1989) claimed that OCB consists of five separate factors. They
included Altruism and Generalized Compliance as suggested by Smith et al. (1983). but they
also included factors labeled Sportsmanship, Courtesy, and Civic Virtue. Sportsmanship is
described as the acceptance of minor inconveniences. Courtesy is offering advice and
respecting other's needs. Civic Virtue represents responsibility and involvement in issues
that affect the entire organization.
Although other researchers have divided OCB into multiple factors, the
conceptualization of OCB as a general factor represents a conceptualizations of OCB that is
consistent with the goal of evaluating a general construct of positive organizational
behaviors. Therefore, no effort will be made to divide OCB into multiple forms in this study.
The past research on OCB relevant to this study consists mainly of research
examining job satisfaction and personality as causes of OCB. Unlike the organizational
withdrawal research cited above, OCB research has, since its conception, had a multi-
behavioral focus. That is, unlike withdrawal, it does not have the one or two high-profile
behaviors (i.e., absenteeism, turnover) that have driven the research. In fact, in his 1988 book
on OCB. Organ (1988) states that a single occurrence of a citizenship behavior is likely to be
trivial. Taken in the aggregate, however, the behaviors can have a significant effect on
organizational performance. This same idea has only recently been proposed in the area of
withdrawal research (Hanisch & Hulin. 1990; Fisher & Locke, 1992).
Recently. Fisher and Locke (1992) addressed the relationships between job
satisfaction and work behavior and provided a review of studies investigating OCB. They
22
concluded that job satisfaction is often a significant predictor of aggregated positive
behaviors. In their review of five studies (both published and unpublished) they report
correlations ranging firom .21 to .54 between OCB and general job satisfaction. Other studies
(e.g. Bateman & Organ, 1983; Motowidlo. 1984; Puffer, 1987; Smith et al.. 1983) have also
consistently reported significant correlations between job satisfaction and OCB.
Bateman and Organ (1983) used a longitudinal design and compared a 30 item
measure of OCB to overall job satisfaction using the JDI (Smith et ai.,1969). They reported
correlations around .40 for both time one and time two data collections. They point out that
correlations of this magnitude are much greater than those traditionally found when
examining job satisfaction and traditional performance criteria.
Smith et al. (1983) also conducted a study examining the relationship between general
job satisfaction (as measured using a semantic differential scale) and OCB. They used a 16
item revised version of the OCB scale used by Bateman and Organ (1983) to assess Altruism
and Compliance. Correlations with job satisfaction were .33 and .29 for Altruism and
Compliance, respectively.
The studies presented above do suggest a consistent relationship between job
satisfaction and OCB. Previously, studies were discussed demonstrating the relationship
between job satisfaction and organizational withdrawal. Given that OCB is an aggregation of
positive non-workrole behaviors and organizational withdrawal is an aggregation of negative
non-workrole behaviors, one could expect similar theoretical relationships. Part of the
purpose of this study was to simultaneously evaluate these two conceptually similar
constructs.
Although conceptually similar, the relationship between positive and negative non-
workrole behaviors is not directly apparent. One possible structure would suggest that they
are opposite ends of a single dimension. This, however, would suggest that it would not be
theoretically possible for employees to engage in both positive and negative non-workrole
behaviors. In other words, if non-workrole behaviors did form a single dimension, they
would be negatively correlated.
A second possible structure for these non-workrole behaviors suggests that the two
behavioral groups are independent. This would allow for an employee to engage in both
positive and negative non-workrole behaviors. It would also suggest that different
antecedents lead to positive non-workrole behaviors than lead to negative non-workrole
behaviors. It would also allow for some antecedents to lead to both. For example, scoring
high on a particular personality trait may lead to both positive non-workrole behaviors and
negative non-workrole behaviors. If non-workrole behaviors existed on a single dimension,
the personality trait would be related to either positive or negative behaviors.
These behaviors have not previously been conceptualized as a single dimension.
Therefore, empirical evidence addressing the specific question of their relation is not
available. Consideration of the research presented above and rational consideration of their
composition suggests that they are not a single construct. As a result, this study considered
the two dimensions to be independent.
The above discussion of attitudes and behaviors at work establishes the first segment
of this study. More specifically, it addresses the relationship between job satisfaction and
non-workrole behaviors. The second major element of this study addresses the relation
between personality and behaviors at work. A review of the literature relevant to this portion
of the study is presented in the next section.
24
CHAPTER 3: LITERATURE REVIEW OF PERSONALITY AND BEHAVIORS AT WORK
Job satisfaction and personality have had similar histories in industrial and
organizational psychology but have mainly followed separate paths. They are similar
because although both seem intuitively important to issues of employee behavior, neither has
produced sufficient empirical evidence to convince the research community of such a claim.
Like job satisfaction, personality has suffered fi:om similar methodological and conceptual
problems that resulted in it losing popularity among researchers in I/O psychology. There
has recently, however, been a renewed interest in personality and its relations to important
organizational variables such as performance and productivity. Improved research
methodology and continued work on the structure and definition of personality has led to
some promising results (see Hogan. 1991). Before discussing the relation between
personality and non-workrole behaviors, it would be useful to briefly review the kinds of
problems that led to the near abandormient of personality in I/O psychology and the
developments that have led to the renewed interest.
The most commonly cited article on personality in the I/O literature is Guion and
Cottier's (1965) review of personality measures used for employee selection. Their stated
goal for the anicle was not to review the entire field of personality, but to summarize the
personality literature that was published in the Journal of Applied Psychology and Personnel
Psychology between 1952 and 1963. In order to be included in the review, an article had to
meet the following restrictions: (1) it had to appear in one of the two target journals in the
12-year period between 1952 and 1963. (2) it had to deal specifically with civilian
employment situations and include an evaluative statement regarding the relation of
personality test data to some measurement of employment success, and (3) it must have used
a personality test that could be found in at least two other studies that met the preceding
restrictions. This last restriction was imposed to eliminate many of the "home-grown"
measures of personality that were often used.
25
The conclusion offered by Guion and Gottier (1965) stated that based on the evidence
from their study, the use of personality in making employment decisions was not
recommended. They did. however, suggest that personality research may have something to
offer I/O psychology, but research up to that point in time did not reveal it. For example.
they refer to the reviewed studies and state "What can be said of these [results] is that they
demonstrate that personality measures have had predictive validity more often than czm be
accounted for simply by chance." (p. 141). Granted, this is not an overly positive assessment,
but it does suggest that the construct does merit further study.
In their conclusions. Guion and Gottier (1965) also refer to an issue, which has since
become known as the "criterion problem." as a possible reason for the poor predictive
validity. In most of the studies reviewed, the criteria consisted of some form of performance
ratings. Guion and Gottier questioned the quality of the ratings and their affect on the results.
They made reference to the fact that most of the criteria used for validity studies consisted of
some form of supervisor rating with little information given regarding what was being rated
or how. Guion and Gottier also commented on the fact that any job descriptions provided
were also quite "sketchy.".
The lack of reliable criterion data automatically precludes the ability to properly judge
the usefulness of the personality construct. Guion and Gottier's (1965) review was conducted
prior to understanding the concept of validity generalization in applied measurement. It was
not until approximately 15 years later that the negative effects of small sample sizes. low
reliability, clerical errors, etc. were generally understood (Schmidt & Hunter. 1981). If one
examines Guion and Gottier's tables that list the attributes of each study reviewed, many of
them had fewer than 100 participants. Also, the most frequent type of criteria listed is
"ratings," which as Guion and Gottier pointed out, were of questionable quality.
It is important to note that Guion and Gottier did a valuable service to the research
community by exposing the poor condition of personality research in I/O psychology. The
26
research community, however, did not respond in such a way as to use the information to
improve the field. Instead, the overall effect of the negative conclusions was a near
abandonment of personality research throughout I/O psychology. The number of studies
conducted directly examining personality and its use in I/O psychology declined greatly
(Bemardin& Bownas, 1985; Fumham. 1992).
Guion and Cottier's (1965) review was just one contributing factor to the lack of
success and subsequent waning popularity of personality in I/O psychology. Another reason
was the focus on measures of abnormal personality. Most traditional personality theories
were developed to explain neurotic or extreme behavior (Hogan, 1991). These theories
originated mainly in clinical psychology, and therefore, were often organized around
psychoanalytic ±eory or other similar doctrines. Examples of these types of measures
include the Minnesota Multiphasic Personality Inventory (MMPI; Hathaway & McKinley.
1943) the F scale (Adomo, Frenkel-Brunswik, Levinson, & Sandford, 1950). and the
Thematic Apperception Test (McClelland, Atkinson, Clark. & Lowell, 1953). The purpose
of these types of tests was to aid in the assignment of people to categories with the intention
of proper diagnosis and treatment. These measures are appropriate for some uses in I/O
psychology. For example, the MMPI is often used to screen individuals who are entrusted
with public safety such as police officers or nuclear power plant operators. These are
positions in which employees with a risk of psychopathology must be carefiilly screened.
However, measures of psychopathology do not necessarily have a connection to the day-to
day activities (e.g.. performance or attendance behaviors) of most workers. Hogan (1991)
makes this point by stating "The absence of neurosis does not necessarily imply the presence
of anything usefiil" (p. 880). As a result, many of the studies using measures developed
around traditional abnormal personality theory were not valid predictors of relevant
organizational variables (Guion and Cottier. 1965).
27
Beyond the research conducted using traditional personality theory, there was another
approach being taken to personality research that focused on the internal structure of
personality. Instead of trying to classify people according to types or categories, researchers
investigated the possible dimensions of personality in which people might differ (e.g..
Cattell. 1965; Fiske, 1949; Guilford. 1975; Norman. 1963). This approach employed a trait
perspective to the conceptualization of personality structure and its research eventually led to
the five-factor model that is generally accepted today (Aamondt, 1996; Barrick & Mount.
1991; Digman, 1990; Hogan. 1991).
The history of the five-factor model (see Digman. 1990 for details) begins with
Klages (1926), a German psychologist, who suggested ±at personality could be understood
by examining language and the words used to describe people. The basic premise states that
langxiage evolves in conjunction with human evolution. Therefore, as people differ on
particular traits, there should be corresponding language that describes these differences.
This idea was implemented by Baumgarten (1933), another German psychologist,
who compiled and examined words describing personality found in the German language.
This same idea was put to use by Allport and Odbert (1933) who developed a list of English
words describing personality. It was finally Cattell (1943, 1946. 1947, 1948) who combined
this lexical concept with factor analysis to identify a number of factors that could be used to
describe differences in personality. His research resulted in a complex taxonomy that
consisted of at least 16 primary factors and eight second-order factors. Cattell's theory was
based on smdies using peer ratings of college students and was considered to be a more
objective description of personality than other theories that preceded it (Digman, 1990).
Fiske (1949) unsuccessfully attempted to replicate Cattell's system. Fiske's
conclusion stated that he was unable to find evidence to suggest more than five identifiable
factors. Similar attempts to replicate Cattell's findings were made by others such as Tupes
(1957) and Tupes and Christal (1961), but Cattell's complex structure was not supported. In
28
fact, the conclusions continued to point in die direction of a five-factor solution. Tupes and
Christal (1961) studied several samples from many backgrounds including males, females,
students, non-students, militar>' personnel, and others. In almost all samples there appeared
to be firm evidence for a five-factor solution.
Other significant studies examining the structure of personality include Borgatta
(1964), Hakel (1974), Norman (1963). and Smith (1967). Of these studies. Norman is most
closely associated with popularizing the five-factor model of personality. In fact, the model
is commonly referred to in the literature as "Norman's Big Five." Norman named the five
factors (I) Emotional Stability, (2) Extroversion or Surgency, (3) Culture, (4) Agreeableness.
and (5) Conscientiousness. The names for each of the factors have varied somewhat from
researcher to researcher. For example, Borgatta (1964) labeled the factors: (1) Emotionality.
(2) Assertiveness, (3) Intelligence, (4) Likeability. and (5) Task Interest. Costa and McCrae
(1985) labeled them: (1) Neuroticism. (2) Extroversion, (3) Openness to Experience. (4)
Agreeableness. and (5) Conscientiousness.
As the various names suggest, there are still some differences in the content of the
factors depending on the theorist. There is, however, general agreement regarding the
number of factors and their basic meanings. The general meaning of each factor will be
described here using descriptions provided by Costa and McCrae (1992). two of the most
prolific researchers of the Big Five and authors of the NEO PI-R which is a personality
inventory designed and developed specifically to measure the five factors of personality. It is
important to note that personality in the context of this study (and as measured by the NEO
PI-R) refers to normal personality.
The first factor is labeled neuroticism. The neuroticism domain deals mainly with the
emotional stability and adjustment of an individual. People who score high on this
dimension are likely to experience negative affect as characterized by fear, sadness, anger,
guilt, etc. Because the focus is on normal personality, even high scores on this dimension
29
can be considered normal and do not necessarily indicate psychiatric problems. Individuals
scoring low on neuroticism can be described as calm, relaxed, and even-tempered.
The second factor is most often referred to as Extroversion but has also been known
as Surgency. (It will be referred to as Extroversion in the remainder of this document.)
People scoring high on the Extroversion factor are often described as outgoing, gregarious,
upbeat, and fiill of energy. Low scorers are often labeled shy, reserved, independent, and
withdrawn. Costa and McCrae (1992) point out that low extroversion scores should be seen
as a "lack of extroversion" as opposed to the "opposite of extroversion." The individual who
is not happy and gregarious does not necessarily have to be sad and depressed.
Openness to experience contains elements such as active imagination, aesthetic
sensitivity, attentiveness to inner feelings, intellectual curiosity, etc. Other labels for this
dimension include intellect or culture. This factor is probably the most difficult to define and
there are still disagreements regarding its meaning (Digman. 1990).
Agreeableness is yet another factor. The agreeable person can be described as
altruistic, sympathetic, helpfiil, and optimistic. The person low on the agreeableness
dimension is likely to be anti-social, difficult to get along with, and possibly narcissistic.
The Conscientiousness factor has received recent attention because of its association
with both work performance and integrity testing (Barrick & Mount. 1993). Costa and
McCrae (1992) suggest that conscientiousness mainly deals with an individual's self-control.
By self-control, they mean the ability to resist impulses as well as the ability to plan.
organize, and carry out tasks. Conscientiousness is also associated with descriptors such as
scrupulous, punctual, and reliable.
These five factors are intended to represent the most basic structure of personality.
Although the factors are considered independent, they are, nonetheless, correlated to some
extent. Costa and McCrae (1992) report correlations among the factors ranging from .02 to
.53 in absolute value. Costa and McCrae also report validity evidence for the five factors
30
derived from a factor analysis indicating a correlated factor structure. Further validity
evidence is presented in the methods section.
As a representation of the most basic structure of personality, and given the criterion
of this study is a limited group of all possible work behaviors, it would be inappropriate to
expect all five factors to be related to the criteria. In fact, the broad application of personality
without regard for its theoretical relationship with criteria has been a common criticism of
personality research (Schneider & Hough. 1995). Therefore, only neuroticism,
agreeableness, and conscientiousness are examined in this study. The definitions of
extroversion and openness to experience do not provide theoretical reasons to expect them to
be related to non-workrole behaviors. Specific expectations about the three personality
factors and non-workrole behaviors are provided below.
Neuroticism is described by Costa and McCrae (1992) as being distressed, nervous,
scomfiil, irrational, and impulsive. It is hypothesized that employees who are high in
neuroticism are more likely to engage in work withdrawal. Similarly, it is expected that
employees who score low on neuroticism are less likely to engage in work withdrawal.
High conscientiousness is described as being punctual, determined, and reliable.
These types of descriptors are theoretically associated with behaviors that make up OCB such
as taking responsibility for initiating changes in your work and staying late to help a co
worker. Accordingly, conscientiousness is expected to be negatively related to work
withdrawal behaviors in this study.
Agreeableness is characterized by being helpful, altruistic, and generally cooperative.
These traits are theoretically associated with positive helping behaviors that make up OCB.
Accordingly, it is hypothesized to be positively related to OCB in this study.
Personality Facets. According to Costa and McCrae (1992) each of the five general
factors are comprised of six subfactors called personality facets. The personality facets that
make up the three personality factors of interest to this study are presented here. Neuroticism
31
consists of facets labeled anxiety, angry hostility, depression, self-consciousness,
impulsiveness, and vulnerability. The agreeableness factor consists of trust-
straightforwardness. altruism, compliance, modesty, and tender-mindedness.
Conscientiousness is composed of competence, order, dutiflilness. achievement, self-
discipline, and deliberation (see Costa & McCrae, 1992)
The same rationale that leads to limiting the number of factors expected to be related
to the criteria of non-workrole behaviors also suggests limiting the nimiber of facets expected
to be related to the criteria. The facets of impulsiveness (from neuroticism), altruism (from
agreeableness), and dutiflilness (from conscientiousness) are of particular interest to this
study because they should be theoretically linked to non-workrole behaviors. Impulsiveness
refers to the desires or cravings that an individual may experience. High scorers find it
difficult to resist temptation. Altruism refers to showing concem for others and a willingness
to help. Dutiflilness refers to following ethical principles and moral obligations. This results
in an individual being considered reliable and dependable.
Based on their definitions, these three personality facets are expected to be more
closely related to the criterion of non-workrole behaviors than the three personality factors
from which they are derived. They are more specific than the personality factors and do not
include extraneous content. Therefore, they are included in the study in order to compare
general and specific levels of criterion-relatedness. Very little organizational research has
been conducted using these particular facets of personality. As a result, little can be said
about their demonstrated relation to the behavioral criteria in this study. Theory, however,
leads to the following expectations. Impulsiveness is positively related to work withdrawal.
Dutifulness is positively related to OCB and negatively related to work withdrawal. .A.nd
Altruism is positively related to OCB. More information is presented about their expected
relationships to non-workrole behaviors in the discussion of hypotheses given at the end of
this section.
32
Criticisms of the Five-Factor Model
Although the five-factor model of personality has received general support from most
personality researchers (Hogan, 1991). diere are some who are not convinced it is the best
way to represent the entire construct of personality. Waller and Ben-Porath (1987) claim that
the five-factor model is not comprehensive because it does not account for other established
theories of personality (e.g.. Murray. 1938). Block (1993) expresses concern regarding its
representativeness because of trait descriptors that have been omitted during its factor
analytic development. For example, he points out that Cattell's (1945) list of 35 variables
that has served as the basis of much of the factor analytic research that led to establishing the
five-factor model, is only 0.1% of the 4505 trait descriptors that were originally compiled by
Allport and Odbert (1936).
In their recent review of the use of personality in I/O psychology, Schneider and
Hough (1995) assess concerns regarding the appropriateness of the five-factor model and
conclude that "It is not time for I/O psychology to embrace the five-factor model of
personality." (p. 84). Their statement does not, however, mean that it should be abandoned.
The five-factor model is a useful structure for personality, and its contributions to personality
research have not yet been fully investigated. Therefore, research should continue on both
the five-factor model as well as other possible conceptualizations of the construct.
It is not the purpose of this study to determine the "truth" about the appropriate
internal structure of the personality construct. The five-factor model is a consistent and
durable representation of personality that is widely accepted and is being used extensively in
several areas including coimseling, clinical, and educational psychology (see Costa and
McCrae, 1992 for details). Because of its dominance in the field, it is important to know as
much as possible about its relationships with other variables. While recent attention has been
directed within I/O psychology toward identifying the role of the five-factor model in
employee selection, performance, and productivity (Schneider & Hough, 1995; Hogan,
1991; Barrick & Mount 1991; Rosse, Miller. & Barnes, 1991; Tett Jackson. & Rothstein.
1991; Barrick & Mount. 1993), research on its relation to other work behaviors such as work
withdrawal or OCB is conspicuously absent.
As was discussed previously, work withdrawal and OCB have been studied as
consequences of job satisfaction. This study also includes personality as an e.xplanatory
variable in the prediction of these non-workrole behaviors. Because both job satisfaction and
personality are investigated as antecedents to non-workrole behavior in this study, it is useful
to first examine how personality and job satisfaction have been shown to be related to each
other in previous studies.
Personality and Job Satisfaction
A review of the literature that directly examines the relationship between personalit\-
and job satisfaction is difficult because of the many conceptualizations of personality. In
fact, few studies in I/O psychology have actually focused specifically on personality and its
relation to job satisfaction. This is at least partially due to the low regard in which
personality has been held by researchers in I/O psychology. There have, however, been
several studies that have examined similar concepts that are closely related to personality-
even if they are not referred to as such. These studies exzonine what is collectively known as
"the dispositional approach to job satisfaction."
The dispositional approach to job satisfaction developed firom the idea that some
people are predisposed to be more satisfied than others regardless of their work situation.
This sparked a flurry of research that aimed to determine the role of disposition in causing
job satisfaction (e.g., Arvey, Bouchard, Segal, & Abraham, 1989; Gerhart, 1987; Judge &
Hulin. 1993; Levin & Stokes. 1989; Staw & Ross, 1985; Schneider & Dachler. 1978; Staw.
Bell, &. Clausen, 1986). Traditionally, most job satisfaction studies examined
extemal/situational influences on job satisfaction such as job characteristics (Hackman &
Oldham, 1976) or social information processing (Salancik & Pfeffer, 1977). The
34
dispositional approach, however, looked to internal influences to explain reported job
satisfaction.
A key piece of evidence regarding the influence of disposition on job satisfaction was
to establish the stability of individuals" levels of satisfaction across time and situations. Staw
and Ross (1985) analyzed reports of satisfaction from over 5,000 men that were part of a
longitudinal study conducted by the Center for Human Resource Research (1977). The
results showed statistically significant correlations across a five-year period. The correlations
of satisfaction over time ranged from .29 to .44. Staw and Ross also reported statistically
significant satisfaction correlations when the employees had changed both occupation and
employer in the five-year period. Correlations here were expectedly lower because of time
and changing jobs, but the correlations were still statistically significant and ranged from .19
to .31.
Judge and Hulin (1993) tested a causal model that predicted a dispositional influence
on job satisfaction and reported that the results do support such a model. Judge and Hulin
predicted that disposition led to subjective well-being (a representation of life satisfaction)
and that the relation between subjective well-being and job satisfaction was reciprocal. The
model also included demographic and other external variables such as alternative
employment opportunities and wages. The path coefficient representing the influence of
disposition on subjective well-being was .58. The path coefficient for subjective well-being
to job satisfaction was .36. Judge and Hulin concluded that their findings provided strong
support for a dispositional basis of job satisfaction.
Contrary to the studies just discussed, other researchers have concluded that
disposition does not play a significant role in determining job satisfaction (e.g., Gerhart,
1987; Newton & Keenan, 1991). Gerhart, like Staw et al., took a longitudinal approach to
comparing dispositional and situational effects on job satisfaction. His conclusions stated
that there was some evidence of the stability of job satisfaction measures across time, but also
35
that situational variables, such as job design, affected job satisfaction to a greater extent than
did dispositional factors. He concluded that the relation between disposition and job
satisfaction was probably not significant enough to be of much concern.
Newton and Keenan (1991) also applied a longitudinal approach and came to a
conclusion similar to Gerhart's (1987). They state that there is evidence that suggests a stable
propensity to report similar levels of satisfaction across time. They point out, however, that
the stability does not necessarily confirm a dispositional cause. They suggest that the
correlations between reports over time may be the result of ongoing stabilitv* across
situations.
Other studies dealing with internal influences of job satisfaction examine the role of
disposition as defined by the traits of positive or negative affectivity (Agho, Mueller. &
Price; 1993; Decker & Borgen, 1993; Fumham Sl ZacherL 1986; Levin & Stokes, 1989; Sah
& Ojha, 1989). The concepts of positive and negative affectivity come fi-om the personality
literature and are more closely related to traditional concepts of personality found outside I/O
psychology. In fact, positive and negative affectivity have been shown to correspond to
extroversion and neuroticism from the five-factor model, respectively (Clark & Watson,
1991). As a result, smdies examining job satisfaction and positive and negative affectivitv-
are also relevant to the relationship between personality and job satisfaction.
Levin and Stokes (1989) investigated the relative effects of negative affectivity and
task design (job characteristics) on reported job satisfaction. They reported that adding
negative affectivity to a regression equation predicting satisfaction with the work itself as
measured by the JDI. improved the variance accounted for above and beyond that accounted
for by task design. The increase in variance accounted for was 4.5% and was statistically
significant. The zero-order correlation reported between negative affectivity and job
satisfaction was -.29.
36
Decker and Borgen (1993) also included measures of job satisfaction in a study
examining the relations between negative affectivity and measures of stress, strain, and
coping at work. They found that negative affectivity accounted for approximately three to
five percent to the variance in job satisfaction beyond the other variables. Contrary to Levin
and Stokes' (1989) who found a similar value, they concluded that the percentage of variance
accounted for by negative affectivity did not add much to the prediction of job satisfaction.
Further discussion of positive and negative affectivity may shed some light on its
relation to job satisfaction. A distinction needs to be made between positive and negative
affectivity as traits (i.e., dispositions) and positive and negative affectivity as states (i.e.,
mood). Tellegen (1985) writes that positive and negative affectivity traits were identified as
higher-order dimensions when factor analyzing responses fi-om scales intended to measure
mood. These higher order factors have come to be considered similar to personality traits
and represent relatively stable aggregates of the responses made on mood scales (see
Tellegen, 1985; Watson & Clark. 1984). Therefore, positive and negative affectivity as traits
represent the tendency to experience positive and negative affect over time. As would be
expected, positive and negative affectivity have been shown to be related to the experience of
positive and negative mood states at work (George, 1989, 1992).
Affective States. Mood at work is conceptually similar to job satisfaction because it
involves how people feel at work. However, mood has been shown to be a distinct element
that is different than traditional conceptu£ilizations of job satisfaction (George. 1989). Job
satisfaction has been shown to consist of a cognitive element and an affective element. Brief
and Roberson (1989) suggest that three popular measures of job satisfaction, the Minnesota
Satisfaction Questiormaire (Weiss. Dawis. England, & Lofquist. 1967), the JDI (Smith et al..
1969). and the Faces Scale (Kunin, 1955), consist mainly of cognitive elements. Mood
therefore, composes an important affective element that is not being completely captured in
these popular measures of job satisfaction.
37
Affective state (mood) has also been shown to be distinct from personality, which is
the other determinant of non-workrole behavior proposed in this study. George (1989. 1991)
has demonstrated that, although positive and negative affective states are partially determined
by positive and negative affectivity (traits), they do not necessarily lead to the same types of
behavior. That is, she found positive affective states to be related to decreased absenteeism
and increased prosocial behavior. However, the trait of positive affectivity was not reported
to be directiy related to either behavior. Positive affective state was reported to correlate -.38
with turnover intentions and -.28 with absenteeism (George, 1989). Negative affective state
correlated . 17 with turnover intentions but only .03 {p >.05) with absenteeism. Correlations
for positive affectivity (trait) were reported as -.01 (p >.05) and -.10 (p >.05) for turnover
intentions and absenteeism, respectively. Negative affectivity (trait) was shown to correlate
.25 and .08 [p >.05) with turnover intentions and absenteeism, respectively.
The theoretical implications of the above results for this study is that affective state is
suggested to have a unique influence on behavior beyond both personality and job
satisfaction. Affect is a temporary, emotional construct that can influence behavior
differentiy than the stable, cognitive constructs of personality and job satisfaction. Therefore,
it was also included as an antecedent of work withdrawal and OCB.
Personality and Non-Workrole Behaviors
Like the personality and job satisfaction literature, there are few studies available that
have directly examined the relation between personality and non-workrole behavior such as
OCB and work withdrawal. However, smdies that examine both personality and OCB are
somewhat more common than studies examining personality and work withdrawal. Many of
the historical reasons associated with personality discussed above also explain the lack of
studies examining its relationship with non-workrole as well. That is. because personality
had fallen out of favor within I/O psychology, it was not investigated in relation to job
satisfaction or non-workrole behaviors. Another reason little research exists in this area is
38
that work withdrawal, as conceptualized in this study, simply has not existed for very long.
Therefore, the review is limited to withdrawal studies that have focused on single behaviors
such as absenteeism and turnover.
Personality and Withdrawal. The problems that have generally kept personality out
of selection research over the years have also kept it out of studies being conducted on
withdrawal behaviors. Weiss and Alder (1984), in their review of personality and
organizational behavior, stated that the nomological networks from the two areas of research
have yet to be combined. Studies have been conducted in the area, but the lack of a unifying
structure for personality research has resulted in many different conceptualizations of
personality. This and the basic lack of research attention on personality in I/O psychology
makes it difficult to draw relevant conclusions from the existing withdrawal literature.
Studies on personality and withdrawal that do exist have generally found that using
personality will improve prediction of withdrawal and withdrawal intentions (e.g., Bemardin.
1977; Jenkins, 1993; Judge. 1993; Kemery, 1991; Mowday & Spencer, 1981). Bemardin
examined the correlations between absenteeism and scores on the following traits; emotional
stability, anxiety, achievement orientation, aggression, independence, self-confidence, and
sociability. He reported that anxiety (similar to neuroticism) and conscientiousness were
predictive of absenteeism. Correlations between anxiety and absenteeism ranged from .21 to
.25. Correlations between conscientiousness and absenteeism ranged from -.40 to .-21. He
concluded that personality variables account for a small, statistically significant amount of
variance in absenteeism. The results, however, are based on small sample sizes (« < 60) and
should be interpreted carefully.
Kemery (1993) also reported that personality predicts withdrawal behavior as defined
by the number of days absent in a six-month period. That interpretation of his results,
however, is questionable. His study examined the relative effects of role conflict, role
ambiguity, and personality as conceptualized by positive and negative affectivity. Using
39
hierarchical regression analyses, Kemery showed that affective disposition contributed to
withdrawal beyond that which was predicted by role conflict and role ambiguity. Although
the results were statistically significant, the additional variance accounted for was very small
7 {R~ = .004). Kemery measured personality by collapsing positive and negative affectivity
into a single bi-polar scale. Tellegen (1985) clearly demonstrated the existence of two
separate dimensions of positive and negative affectivity. Combining the two activity
dimensions may have artificially weakened Kemery's observed relationship with
absenteeism.
Mowday and Spencer (1981) also showed that personality could predict absence
behavior and turnover. Their study too, reported that the personality regression coefficients
were statistically significant, but the variance accounted for was quite small {R = .02).
Mowday and Spencer used scales measuring need for achievement among the participants.
Achievement is considered to be one element that makes up the conscientiousness scale in
Costa and McCrae's NEO PI-R (1992). The reliability of the scale used by Mowday and
Spencer to measure personality was .43. They defended the low reliability of the scale by
stating that measures of complex personality traits seldom attain the levels of internal
consistency expected of other measures. Nonetheless, their results should be interpreted
carefully.
Ferris, Youngblood, and Yates (1985) used personality to predict training
performance and withdrawal in a sample of flight attendants. The study used four second-
order factors from the 16PF, a measure of 16 factors of personality (Cattell. Eber, &
Tatsuoka, 1970). The factors were Extroversion, Anxiety, Corteria (which is described as
"tough poise"), and Independence. The Anxiety scale, which is similar to neuroticism as
measured by the NEO PI-R (Costa & McCrae, 1992), was the only scale shown to be
negatively correlated with attendance (measured as the amount of sick leave not used in a
one-year period; r = -.34). Turnover did not show a statistically significant correlation with
40
any of the four scales. Methodological concerns regarding this study such as non-normally
distributed behaviors hinder the interpretation of the results.
The relatively frequent occunence of studies reporting statistical significance but
small effect sizes does not paint an optimistic picture regarding the ability of personality to
predict individual withdrawal behaviors. However, it is difficult to draw any conclusions
from this literature because of the lack of coherent methods and the incongruency of
personality as a general measure and the use of isolated instances of organizational
withdrawal behavior. Much like Roznowski and Hanisch (1990) demonstrated with the
relation between job satisfaction and aggregated measures of organizational withdrawal, it is
expected that there is much to be gained by using personality and aggregated measures of
organizational withdrawal to leam about their relationship.
Personality and OCB. The situation regarding personality and OCB is somewhat
different than the situation regarding personality and withdrawal. OCB research was initiated
through the application of ideas and findings from the study of prosocial behavior in social
psychology. Because of its origins in prosocial behavior, OCB was more closely tied to
explanations involving personality. Personality had been actively researched as an
antecedent to prosocial behavior by social psychologists (e.g., Gergen, Gergen. & Meter.
1972; Underwood & Moore, 1982). Therefore, it logicjilly followed that personality would
also be considered as an antecedent to OCB. However, the personality research conducted
within social psychology suffered from many of the same concepmal problems regarding the
use of personality in I/O psychology. In an early review of the prosocial literature, Krebs
(1970) stated that studies investigating prosocial behavior and personality "were plagued with
difficulties" (p. 298). Despite the mixed results described in his review, Krebs did claim that
there was some evidence suggesting that people low in neuroticism and high in extroversion
were more likely to engage in prosocial behavior than those people who were not.
41
In more recent research, Bonnan and Motowidlo (1993) present research suggesting
that personality is most likely relevant to what they refer to as "contextual performance" as
opposed to traditional task performance. They stated that employees who are effective in
contextual performance are likely to be empathic. extroverted, well adjusted, cheerful, and
achievement oriented. Data reported by Hough and Schneider (1995) supports this
proposition. They present correlations for personality with their measures of contextual
performance which they labeled "commendable behavior" and "law abiding behavior" of .16
and .39, respectively. These values become more meaningful when compared with the
correlation reported for personality and task performance (r = .08).
Organ (1994) suggested that the agreeableness and conscientiousness factors of
personality are very similar to the altruism and compliance factors of OCB. Organ and LingI
(1995) as cited in Organ (1994), studied the agreeableness and conscientiousness components
of the Big Five and found moderate correlations between personality and OCB.
Agreeableness correlated .20 with the altruism factor of OCB. The correlation between
conscientiousness and the compliance OCB factor was .30. Organ suggested that the
correlations between personality and OCB could be improved by using measures of
personality that were even more directly related to OCB. He recommended investigating the
more specific facets of the Big Five that have been identified by Costa and McCrae (1992)
because OCB like many behaviors of interest to psychologists, is likely to be more complex
than the simple structure provided by each of the Big Five factors of personality. This view
is congruent with other authors writing recent reviews regarding the use of more construct-
relevant personality measures in I/O psychology (e.g., Guion, 1991; Hogan, 1991: Schneider
& Hough, 1995).
Overview
The literature review indicates that personality, affect, and job satisfaction have not
shown consistent relationships with traditional workrole behavior (e.g., performance) or
42
traditional, single-incident withdrawal behavior (e.g.. absenteeism, turnover). It does,
however, suggest that personality, affect, and job satisfaction may be related to non-workrole
behavioral families such as OCB and work withdrawal. The study presented here examines
the relations among personality, affect, job satisfaction, and non-workrole behaviors in a
causal modeling framework. Using two series of nested structural models, a priori
comparisons are made to investigate the contribution of job satisfaction, positive and
negative affectivity. and personality to the prediction of OCB and work withdrawal. In order
to maximize statistical power and minimize possible problems associated with a small
sample size, the theoretical models to be tested are organized in two parallel series. The
models are the same in both series except for the specification of personality. In the first
series of models, personality factors are specified. In the second series of models.
personality facets are specified.
Studies of personality have included various forms and conceptualizations of the
construct. Some studies have conceptualized personality very broadly within the content of
personality without concern for the smdy's criterion. For example, George (1990) uses the
very general positive and negative affectivity to represent personality. Others have
conceptualized personality more precisely. For example, Mowday and Spencer (1981) used
"manifest needs" as a very specific and somewhat esoteric conceptualization of personality.
In the study presented here, the level of personality measurement is varied but in a
systematic way. Here, personality has been conceptualized in a general form using
personality factors as defined in the five-factor theory of personality and in a more specific
form using personality facets that have been identified as lower-order factors within the five-
factor structure. This creates a situation similar to that encountered with the congruence of
level of measurement between attitudes and behaviors. It is different, however, because in
the relationship between personality and non-work role behaviors, it is the predictor that is
believed to be too broad to be appropriately compared to the behavioral criteria. While work
43
withdrawal and OCB are broad behavioral families, the personality factors each contain
elements that are not expected to be related to the criteria in this study. Consequently, it is
hypothesized that the personality facets are more congruent with the criteria of work
withdrawal and OCB in terms of their level of measurement, than the more inclusive
personality factors.
As was stated previously, Costa and McCrea (1992) have determined that each of the
Big Five factors of personality contain six lower-order personality factors referred to as
personality facets. The existence of these lower-order factors provides a simple, unified
structure by which to include differing levels of measurement for the construct of personality.
Similar to the personality factors, the uiclusion of the particular personality facets in
this study is based on consideration of their theoretical relationships with the non-workxole
criterion behaviors. Because there have been no prior studies directly examining the
relationship of the personality facets to organizational withdrawal or OCB variables, the
inclusion of the personality facets is based on consideration of their theoretical definition (as
given by Costa and McCrae, 1992) and existing information about the higher-order factors to
which they belong. Based on the existing literature and consideration of theoretical
relationships, the following models were developed to be tested in this study. The series of
models containing personality factors (series 1) is described first.
Figure 2 presents Model 1 of the first series which is the null model. It includes no
causal relationships, and it serves as a baseline for model fit statistics to which the successive
models are compared. It only includes the correlational paths between the exogenous
variables to allow for the expected intercorrelations between them. The exogenous variables
are allowed to be correlated in the models to account for multicolinearity among the predictor
variables. By first testing the null model with only the intercorrelations among exogenous
variables allowed to be estimated, the subsequent models will provide a more accurate view
of the causal influence of each of the added variables. In other words, if the hypothesized
Neurol icism
Factor
Agreeableness
l-actor
Negative
Affect
Positive
Affect
Conscientiousness
w i'actor J
Figure 2. Model 1: The Null Model 1-or Series 1
Work
Withdrawal
Organizational Citizenship
Behavior
45
models were compared to the true null model (i.e.. a model with no relationships at all), the
increase in goodness of fit statistics would be somewhat inflated as a result of allowing the
exogenous variables to intercorrelate in the hypothesized models but not the null model.
Figure 3 presents Model 2 which builds on Model I. Model 2 includes the
hypothesized relationships showing job satisfaction causing work withdrawal and OCB. Bv
adding only job satisfaction to the model, this step will demonstrate the influence of job
satisfaction on the two criterion variables. Subsequent models which include additional
variables can then be compared to this model to demonstrate their relative contribution to the
prediction of the criteria. As was discussed in the literature review, the relationship between
job satisfaction and the two criterion variables has been demonstrated frequently in previous
studies. Therefore, it is expected that there will be a substantial improvement in the fit of the
model by adding these causal paths.
Figure 4 presents Model 3 which continues to build on Model 2. In this model, causal
paths are added from positive affect to OCB and from negative affect to work withdrawal.
Positive and negative affect have been shown to have distinct effects on withdrawal
behaviors and OCB beyond that of job satisfaction or personality. Accordingly, an
improvement in model fit is expected to coincide with freeing these causal paths.
Figure 5 presents Model 4 which is expected to be the best-fitting model in this series. It
adds to Model 3 by freeing four causal paths. Those paths are from neuroticism to work
withdrawal, from conscientiousness to both work withdrawal and OCB, and from
agreeableness to OCB. The path from neuroticism to work withdrawal is included because
employees who are nervous, anxious, and impulsive are hypothesized to engage in work
withdrawal behaviors more than employees who are not. The paths from conscientiousness
to both work withdrawal and OCB are explained by the expected relation between the
variables based on their definitions. Similarly, the path from agreeableness to OCB is
Neuroticism
Factor
Conscicnliousnes?
, Factor j Work
Withdrawal
Agrecableness
Factor
Negative
Affect
Organizational Citizenship
Behavior Job
Satisfaction
Positive
AI feet
figure 3. Model 2: Testing Job Salisfaclion Tor Series 1
Ncurolicism
Tactoi
Agreeableness
FaOor
Negative
Affect
Job
Satisfaction
Conscientiousness
i'actor y
Positive
Affect
Figure 4. Model 3; Testing Positive and Negative Affect l-or Series
Work
Withdrawal
Organizational Citizenship
Behavior
Neuroticism
Factor
Conscientiousness
factor
Agrecableness
Factor
Negative
Affect
Job
Satisfaction
Positive
Affect
I'igure 5. Model 4: Testing i'ersonality I'or Series I
Work
Withdrawal
Organizational Citizenship
Behavior
4^ 00
49
included because of the theme of being helpful and concerned that is common to both
concepts.
The second series of models follows the same progression of series 1 except series 2
uses the personality facets of impulsiveness, dutifulness, and altruism in place of the
personality factors neuroticism, agreeableness, and conscientiousness. Each of the
personality facets is conceptually part of the more general factor it is replacing. Therefore,
impulsiveness is a subset of neuroticism, dutifulness is a subset of conscientiousness, and
altruism is a subset of agreeableness. The same rationale was used for their inclusion as was
used for the inclusion of the personality factors. The difference, however, is the definition of
each facet and its expected relation to the criteria.
Figtires 6-8 show the same progression demonstrated in series 1 of establishing the
relationships between the criteria and job satisfaction and affect. Figure 9 shows model 4
which illustrates the relationships for the personality facets to work withdrawal and OCB.
Similar to the first series of models, the rationale for including the specified causal paths is
based on the dieoretical relationships between the specific facets and the criterion variables.
Model 4 in the second series of models is proposed as the best fitting of all eight
models. .A.s was indicated in the literature review above, recent success using personality in
the prediction of work behavior has been realized in separate instances by focusing on non-
workrole behaviors, construct-related conceptualizations of personality, and positive and
negative affective states. This model incorporates each of these elements plus job
satisfaction in a way that is expected to accurately represent their empirical relationships.
Impulsiveness I'acet
Altruism Facet
Negative
Affect
Positive
Affect
iMgurc 6. Model 1: The Null Model l-'ot St-ries 2.
Work
Withdrawal
^ o / Organizational \ I Citizenship )
Behavior V
Impulsiveness I'aci't
Dutifuliiess Facet
Altruism Facet
Negative
Affect
Job
Satisfaction
Positive
Affect
Figure 7. Model 2: Testing .lob Satisraction lH)r Scries 2.
Ul
Impulsiveness luicet
Duti fulness Facet
Work
Withdrawal
Altruism Facet
Negative
Affect
Organizational Citizenship
Behavior Job
Satisfaction
Positive
Affect
I'igure 8. Model 3: Testing Positi\e and Negative AlTecl l-'or Scries 2.
Impulsiveness I'acfi
Dutifulncss Facet
Altruism Facet
Negative
Affect
Job
Satisfaction
Positive
Affect
iMgure 9. Model 4: Testing Personality I'or Series 2.
Work
Withdrawal
U)
Organizational Citizenship
Behavior
54
CHAPTER 4: METHOD
Sample
Participants in the study were employees from two organizations providing health
care to disabled adults in the rural mid-west. Questionnaires were distributed to employees
in both organizations. Overall, 605 employees received questiormaires; 240 in Organization
1 and 365 in Organization 2. One hundred and thirty-four questionnaires (56%) were
returned from Organization 1 and 191 questionnaires (52%) were returned from Organization
2. Of the 325 questiormaires that were returned, 12 were removed from the sample because
of excessive missing data. A questiomiaire was considered to have excessive missing data if
any of the relevant scales were missing values for 25% or more of their items. For example,
if a 12-item personality factor scale had fewer than nine valid items, the entire case was
removed from the data set. A mean replacement strategy was used for those cases having
missing data that were not considered excessive. Of the remaining 313 cases. 97 had at least
one missing value that was replaced with the variable's sample mean.
After dealing with missing values, the analyses were conducted on a final sample of
313 cases. Several categories of demographic information were collected in the
questionnaire. Appendix A provides the questions and possible response options for the
demographic data. As might be expected in a residential healthcare industry, the majority of
the sample was female and made up 81 % of the sample. The respondents were mostly
married (67%), ftill-time employees (72%) who ranged in age from 16 to over 65. The most
common age categories were ages 36 to 41 and 42 to 47. Almost the entire sample had an
education level of high school or above (96%) and nearly 38% reported having at least a
college degree (AA. BA. BS). Salaries for the sample were moderate to low with 88%
reporting armual salaries of $25,000 or below.
The sample size of 313 is considered adequate for structural equation modeling based
on conventions established by other researchers. Anderson and Gerbing (1988) stated that a
sample size of 150 is usually large enough to obtain a proper solution and avoid non-
convergence if there are at least three indicators per variable. Unlike exploratory factor
analysis or regression analysis, sample size concems in structural equation modeling do not
focus on the ratio of variables to cases. Instead, they focus on the ratio of estimated
parameters to cases. As a benchmark standard. Bentler and Chou (1987) recommend a
minimum of five cases for each estimated parameter. For this study, this issue was of
greatest concern in the context of running the measurement models; where the greatest
number of parameters (76) were being estimated. Using 313 cases to estimate 76 parameters
produces a parameter to case ratio of just over four to one. Given that the five to one ratio is
a recommended standard, there were three indicators per construct, and no problems of
convergence were encountered (the measurement models reached convergence in 15
iterations), the sample size was determined to be adequate.
Measures
Each of the measurement scales is presented as an appendix to this paper. In the
appendices, the items are organized by content. In the questionnaire mailed to participants,
many of the items were presented in random order to alleviate possible response sets. For
example, conscientiousness items were mixed with neuroticism items and work withdrawal
items were mixed with OCB items.
Job satisfaction. Job satisfaction was assessed using the Job Descriptive Index (JDI;
Smith et al.. 1969: Roznowski, 1989: see Appendix B). The JDI is a measure that assesses
five distinct facets of general job satisfaction. The facets of satisfaction measured by the JDI
are work satisfaction, pay and benefits satisfaction, co-worker satisfaction, supervisor
satisfaction, and promotion satisfaction. Parsons and Hulin (1982) demonstrated that the
facet scales do share a communality that suggests a second-order general factor that runs
through each of the facets. Therefore, the summed scales also serve as a measure of overall
job satisfaction. The psychometric qualit>' of the JDI is excellent, and it is considered to be
56
one of the best measures of job satisfaction available (Roznowski, 1989). The validity of the
instrument has been thoroughly researched using multiple methods including factor analysis,
item response theory, and causal modeling (Hanisch. 1992; Parsons &. Hulin, 1982;
Roznowski. 1989). It has also been shown to be valid across many different cultural groups
(e.g.. Hulin & Mayer, 1986; Smith. Balzer. Brarmick, Eggleston. Gibson. Ironson, Josephson.
Paul, Reilly, & Whalen. 1987). Reliability estimates for the five subscales are reported to
range from approximately .80 to .90 (Roznowski. 1989). For this study, the five subscales
were by summed to provide an overall measure of general job satisfaction.
Personality. Neuroticism, conscientiousness, and agreeableness were assessed using
the NEO Five-Factor Inventory (NEO-FFI;Costa & McCrae, 1992; see Appendix C). The
NEO-FFI is a shortened version of the more comprehensive NEO PI-R, a personality-
inventory designed specifically to measure the five-factor structure of personality. Schneider
and Hough state that the NEO PI-R is "the most frequently used and best researched measure
of the five-factor model" (1995, p. 80). The NEO PI-R is a measure of normal adult
personality that consists of five domain scales: Neuroticism (N), extroversion (E), openness
to experience (0). agreeableness (A), and conscientiousness (C). The domain scales in the
full NEO PI-R are made up of 48 items each. Each domain scale contains six facet scales
that measure more specific aspects of each domain. The NEO PI-R was created using a
combination of both rational (deductive) and factor analytical (inductive) approaches to scale
development. Coefficient alphas for the general scales measuring the five domains are .92.
.89, .87, .86 and .90 for N, E, O, A, and C, respectively (Costa & McCrae, 1992).
Because of limits imposed on the length of the questionnaire, a short form of the NEO
PI-R, the NEO-FFI was used in this study. The NEO-FFI is composed of a subset of 60
items (12 items per scale) taken from the NEO PI-R. According to Costa and McCrae
(1992), items for the NEO-FFI were selected by examining factor loadings of the NEO PI-R
on the domain scales. Those items with the highest positive or negative loadings were
57
selected for the shortened scales. A few substitutions were then made to ensure content
diversity and response direction. The correlations between the scales in the NEO-FFI and the
more comprehensive NEO PI-R are .92. .90. .91, .77, and .87 for the N. E, O. A. and C
domains, respectively. Reliability estimates (coefficient alpha) reported by Costa and
McCrae for each of the scales of the NEO-FFI are .86, .77, .73, .68, .and .81 for the N, E. O,
A, and C domains, respectively. These coefficients are expectedly lower than the estimates
for the much longer NEO PI-R, but are a necessary sacrifice because of time and space limits
on the questionnaire. Three-month test-retest reliabilities do show that the NEO-FFI scores
are stable over time. These coefficients were reported by Costa and McCrae to be .79. .79.
.80. .75. and .83 for N, E. O. A. and C, respectively.
Validity evidence for the NEO PI-R (long form) is extensive; it is one of the most
researched measures of the five-factor model. One form of validity evidence is demonstrated
by the internal structure of the measure accurately corresponding to the five factors it is
intended to represent. Costa. McCrae, and Dye (1991) used factor analysis to analyze the
240 items in the NEO PI-R. The result was a factor structure consistent with the Big Five.
External validation has been demonstrated by relating the NEO Pl-R to other
personality measures. For example, the NEO PI-R has demonstrated convergent and
discriminant validity with other scales such as the Eysenck Personality Inventory' (Eysenck &.
Eysenck, 1964) and the Personality Research Form (Jackson, 1984). Construct validity has
been demonstrated by showing the relation between NEO PI-R scores and various personality
constructs such as psychological well being, coping, and interpersonal traits. Details on the
development and validation of the NEO PI-R can be found in the NEO PI-R Professional
Manual (Costa & McCrae, 1992).
Unfortunately, as is often the case with shortened versions of psychological measures,
there is not as much information available for the NEO-FFI as there is for the NEO PI-R.
Costa and McCrae (1992) point out, however, that as a subset of the NEO PI-R, the NEO-FFI
58
will carry a portion of the demonstrated validity of the longer version. They state that the
NEO-FFI scores capture about 85% of the variance accounted for by the full domain scales
on the NEO PI-R.
Costa and McCrae (1992) present convergent and discriminant correlations between
the NEO-FFI and an adjective checklist measure of the Big Five that had been completed
three years earlier. The convergent correlations ranged from .56 to .62 and none of the
divergent correlations were in excess of .20. These values seem acceptable given the
different methods of assessment and the three year period between measurements (Costa &
McCrae, 1992).
The response format for the NEO-FFI is a five-point scale from definitely
false/strongly disagree to definitely true/strongly agree. Participants use this format to
respond to questions such as I am not a worrier, I am not a very methodical person, and I like
to be where the action is.
A validity issue that has been associated v^dth personality assessment is the possibility
of participants faking their responses to manipulate how they are perceived (Hogan. 1991).
Although it has been shown that it is possible to fake responses to obtain a more desirable
score or profile, there is reason to believe that it is not a widespread problem (Hough. Eaton.
Duimette. Kamp. & McCloy, 1990). Hough et al. (1991) conducted a study in which 245
soldiers completed personality inventories twice in one of the following conditions: (1) Fake
good-Honest, (2) Honest-Fake good. (3) Fake bad-Honest, or (4) Honest-Fake bad. Fake
good meant answering the questions to try and get into the army. Fake bad meant answering
the questions as if trying to avoid getting into the army. The Honest condition meant
answering the questions truthfully. These scores were then compared to the scores of army
applicants who had recently gone through testing. Hough et al.'s conclusions stated that
when instructed, soldiers could distort their scores on the personality inventory. However,
comparing the soldiers' scores to the actual applicants' scores showed that the applicants'
59
scores resembled the scores of the soldiers in the Honest condition. Therefore. Hough et al.,
concluded that the applicants did not distort their responses.
Similar studies to Hough et. al's have been conducted by Orpen (1971) and
Abrahams, Neumann, and Githens (1971) with non-military samples resulting in similar
findings. Orpen compared personality scores of applicants for a clerical position to a group
of "matched" students who were told to distort their scores. Abrahams et al. compared "true"
and "faked" student scores on a vocational interest inventory to actual interest inventory
scores taken under normal circumstances. Both Orpen and Abrahams et al. reported diat
scores can be faked, but comparisons between scores from instructed "true" conditions were
virtually identical to scores taken from "normal" testing situations.
One approach that has been taken in the past to deal with intentional distortion on
self-report measures is the use of "lie" or "social desirability" scales. These scales include
items that indicate the respondent's use of response sets to distort his or her perceived image.
Costa and McCrae (1992), however, concluded that the potential benefit gained by including
"lie" or "social desirability" scales does not outweigh the interference that can be caused by
including such scales. They, therefore, do not recommend their use with the NEO PI-R or
NEO-FFI. Tellegen (1985) agrees with this approach and states that "more important than...
detection efforts are attempts to establish a relationship of trust and functional collaboration
with the respondent" (p. 683).
In line with Tellegen's statement, an effort was made to establish trust with the
participants, and to convince them to provide honest, accurate responses for this project.
Participants were informed that they were participating in a research project and not a form of
employee evaluation. They were also assured of confidentiality at every possible
opportunity. The importance of honest, accurate answers was explained to them clearly in
the survey instructions. The researcher also met with a portion of the employees to explain
the project in person and answer any of their questions. Organizational managers were also
60
involved showing support for the project and assuring the participants that the individual
responses would only be seen by the researcher.
Construct-Related Personality Facets. The personality facet scales used as specific
measures of personality were taken from the NEO PI-R long form (see Appendix D). The
specific facet scales were impulsiveness (from the more general factor neuroticism). altruism
(from the agreeableness factor), and dutifulness (from the conscientiousness factor). Each
facet scale consists of eight items and uses the same five-point response scale as the NEO-
FFL Coefficient alphas reported by Costa and McCrae (1992) for each of the facet scales
were .70, .75. and .62 for impulsiveness, altruism, and dutifulness, respectively. Being
subscales of a broad measure of personality, it is expected that the internal consistency
coefficients will be lower than if the scales were developed to measure the facets specifically.
Ideally, the coefficient for the dutifulness factor in particular would be higher than it is.
However, due to the direct relevance of its items to the non-workrole behaviors of interest
(e.g., 1 try to perform all the tasks assigned to me conscientiously, I'd really have to be sick
before I'd miss a day of work), it was included in the questionnaire.
Non-Workrole Behavior As was stated previously, the term non-workrole behavior is
used to refer to both work withdrawal and organizational citizenship behaviors. Although
conceptually different, the two constructs are similar because they are both peripheral to
traditional work performance. The measures of work withdrawal and organizational
citizenship behaviors were presented together in the questionnaire to provide balance
between the positive and negative nature of the questions (see appendices E and F). There
was a total of 41 non-workrole behavior items included in the questioimaire (22 work
withdrawal items and 19 organizational citizenship items). There were two sections within
the questionnaire where non-workrole items were presented. One section consisted of seven
items (five organizational citizenship and two work withdrawal) in a seven-point strongly
agree to strongly disagree format. The second section consisted of 34 items (14
61
organizational citizenship and 20 work withdrawal) presented in a behavioral frequency
format from Never to More than once a week.
Work withdrawal was measured using self-report scales asking the participants about
their instances of behavioral and psychological withdrawal. Multiple items were combined
across behaviors and cognitions representing the removal of one's self from one's work.
Examples of these items include Daydreaming while I should be working. Making excuses to
leave the work area, and Taking frequent or long coffee or lunch breaks. Respondents were
asked to indicate the frequency of each behavior's occurrence using the following eight-point
scale: l=never, 2=maybe once a year. 3=two or three times a year, 4=nearly every month.
5=about once a month, 6=more than once a month, 7=once a week, and 8=more than once a
week.
The work withdrawal items are based on the scales originally used by Roznowski.
Miller, and Rosse (1990) as well as by Hanisch and Hulin (1990) and Roznowski and
Hanisch (1990). The scales are altered slightly from study to smdy (or from sample to
sample) in order to tailor the behaviors to be relevant to a particular sample and organization.
Accordingly, the work withdrawal scale used here included items that relate to jobs in
general as well as items that are meant to be related to the jobs included in this specific
sample.
Before analyses were conducted using the work withdrawal scale, inter-item
correlations and descriptive statistics were examined to evaluate each item's relevance to the
scale. Based on this examination, the item Drinking alcohol or using drugs before coming to
work was removed. The item had a very low item-scale correlation (r = .02) and very little
variance (only two of the 313 participants gave a response other than Never).
Because of the breadth of behaviors intentionally included in the scale, traditional
internal consistency estimates of reliability are expectedly lower than normally found for
homogeneous scales. Hanisch and Hulin (1991) have reported coefficient alphas for work
62
withdrawal scales to be in the range of .51 to .62. Work withdrawal items have also been
shown to cluster together through factor analysis and causal modeling (Hanisch & Hulin.
1990; 1991).
Organizational citizenship behavior was assessed using items based on the scale
developed by Smith et al. (1983). The negatively worded OCB items from Smith et al.'s
measure are more directly related to the work withdrawal construct than to OCB. Therefore,
only positively worded items were used in the OCB measure. Also, some of the Smith et al.
items reflected behaviors that are considered workrole behaviors as opposed to non-workrole
behaviors. These items were not included.
To be more consistent with the measures of work withdrawal in this study, the items
were changed from an objective rater/evaluation point of view to a self-report/frequency of
occurrence point of view. That is, the items were altered to match the employee's perspective
and to match the frequency response scale that is used in the assessment of work withdrawal.
For example, an original OCB item read "Assists me with my duties." It was changed to
"assisting my supervisor with his/her duties." Like work withdrawal, the response scale asks
participants to indicate how frequently they have engaged in the activity in the last 12
months.
Inter-item correlations and descriptive statistics were examined to evaluate each of the
OCB items in the scale. This examination resulted in the removal of two items from the
OCB scale. The items were "I take fewer days off than other employees" and "I give
advanced notice if I'm unable to come to work." These items had low item to total scale
correlations (.05 and .13. respectively).
Affective State. Affective state was measured using the Positive and Negative Affect
Scales (PANAS; Watson, Clark, &. Tellegen, 1988). The PANAS consists of two 10-item
scales which measure positive and negative affect (as a state, not a trait). Each scale consists
of 10 mood descriptors that represent either positive or negative affect (see Appendix G).
63
Participants respond to each descriptor using the following five-point scale: I = very slightly
or not at all, 2 = a little. 3 = moderately. 4 = quite a bit. and 5 = extremely. The PANAS
scales were developed to be used with a range of time references from "Moment (you feel
this way right now, that is, at the present moment)" to "General (you generally feel this way.
that is, how you feel on the average)." Watson et al. (1988) provide reliability and validit\-
information for each of the six time period references. Coefficient alpha for the positive
affect scale ranged from .86 to .90. Coefficient alpha for the negative affect scale ranged
from .84 to .87.
Validity for the scales has been demonstrated through factor analysis as well as
correlations with other measures of mood. Factor analysis demonstrated that the two scales
are orthogonal (Watson et al., 1988). Factor loadings for positive affect ranged from .52 to
.75. Factor loadings for negative affect ranged from .52 to .74. Correlations with other
theoretically similar measures ranged from .50 to .94. Correlations with theoretically
dissimilar measures ranged from -.04 to -.43.
The time frame used for the PANAS scales in this study was the "week" distinction.
The issue of what time frame to use when measuring mood is difficult because, as the time
frame increases, the mood (state) measure becomes more of a trait measure. However, if the
time frame is too short, its relevance to behavior that occurs over time becomes questionable.
By using a shorter time frame, a case can be made regarding mood at the time of reporting
behavior. For the purposes of this study, a time specification was necessary so the affective
state measure was distinct from personality. Based on this consideration, one week was
selected. Other research has also supported this time frame includmg work by George and
her colleagues (1989; Brief, Burke, George, Robinson, & Webster. 1988; George &
Bettenhausen. 1990) in past studies of mood and work behavior.
64
Procedures
Questionnaires were distributed to employees with their paychecks either at the
worlcpiace or through the mail. Participation was completely voluntary. Employees also
received a letter explaining the purpose of the project the extent of their expected
involvement, an assurance of confidentiality, and a reminder that their participation was
voluntary (see Appendix H). A cash lottery was offered as an incentive for the employees to
participate. Employees who completed and returned their questionnaire were entered in a
drawing for $100. The completed questionnaires were retumed via the U.S. Postal Service in
pre-paid, pre-addressed envelopes mailed directly to the researcher. This ensured
confidentiality because completed questionnaires were never seen by die employees'
superiors.
Data Preparation
To ensure the quality of the data, the questionnaires were entered twice: this resulted
in two separate data sets. The two data sets were then compared to each other and examined
for discrepancies. The assumption was that data entry errors would not produce the same
incorrect values in both data sets. Variables in the data sets that showed different values were
checked against the original questionnaire and corrected.
Analyses
Structural equation modeling was implemented using LISREL 8 (Joreskog &
Sorbom. 1993) in a two-step modeling procedure (Anderson & Gerbing, 1988). This
procedure separately evaluates a measurement model and a structural model. The
measurement model links the observed variables to the latent traits. The structural model
indicates the relations among the latent traits. Anderson and Gerbing recommend the use of
the two-stage modeling procedure because they claim that it provides less biased parameter
estimates than a one-step procedure.
65
In the two-step modeling procedure, the quality of the measurement model is
evaluated before the structural model. This allows for a more knowledgeable evaluation
regarding the quality of the structural model than would be possible using a one-step
procedure. For example, if the fit statistics resulting from a simultaneous analysis suggest a
poor model fit. one conclusion might be that the structural model was misspecified. This
misspecification would then be attributed to poor theory. An alternative explanation,
however, would suggest that the poor quality of measurement in the manifest (measurement)
model caused the poor model fit. If the measurement model and the structural model are
analyzed in a single step, there is no way of knowing which conclusion is correct. This
problem is avoided by using the two-step approach.
It should be noted that estimating the measurement model and structural model
separately can give the appearance that the procedures have more statistical power than a
one-step procedure because fewer parameters are being estimated in each model. It is
important to realize, however, that the same total number of parameters are being estimated
in both one-step and two-step procedures. Therefore, the statistical power of the of the entire
two-step procedure is not changed relative to a one-step procedure.
Measurement Models. As was stated previously, the structural equation modeling
was conducted using two series of nested models. The two series were identical except for
the assessment of the personality variables. The first series used the more general personality
factors, and the second series used the more specific personality facets. The same procedures
were used for testing both series and they are described below.
The measurement model for each series was operationalized by creating three parallel
subscales to serve as indicators for each of the latent constructs in the study. The indicators
were created by assigning each item from a given scale to one of three subscales to serve as
manifest indicators. In order to ensure equality among the three indicators in a given set.
consideration was given to the content and response value for each item. Items were
66
distributed to indicators according to their mean value across all cases. For example, to
create three parallel indicators for the agreeableness subscale of the NEO-FFI, the 12 items
were listed in descending order based on their mean response values. The list was then
divided into three groups (i.e., high group, medium group, and low group) of four items each.
The indicators were then created by randomly assigning items to the new subscales from the
high, medium, and low response value groups. Because there were 12 items, the distribution
of high, medium, and low items was not even. That is. after assigning one item from each
category to each new indicator subscale, one high, one medium, and one low item still
remained. These items were assigned to subscales with consideration for the values of the
items aheady assigned. An attempt was made to have the final item be in balance with the
existing items. For example, if a subscale was assigned the lowest of the low values, it
would receive the remaining high value item. Similarly, if a subscale was assigned the
highest of the high values, it would receive the remaining low value item. This was done to
help ensure an equal distribution of item values within each subscale and ensure the parallel
nature of each subscale in the sense of classical test theory.
The procedure of creating three parallel manifest indicators was followed for each of
the constructs in the models. However, creation of the general job satisfaction indicators
warrants further discussion because of the Job Descriptive Index's (Smith et.al.. 1969) multi-
factor composition. Because the JDI was being used as a measure of general job satisfaction,
each of the five underlying factors needed to be equally represented in each of the indicators.
Therefore, the procedure for balancing the responses within each indicator was performed for
each of the five subscales. That is, three indicators were created for each of the five job
satisfaction subscales and then the five subscale indicators were combined to create one of
the diree general job satisfaction indicators. Table 2 contains the means, standard deviations,
and number of items for each of the manifest indicators. Correlations among the manifest
indicators are presented in Appendix I.
67
The means within each set of manifest indicators were within a range of two or three
points of each other. This would suggest that the items within each indicator were grouped
successfully to create three equal measures. There were, however, a few exceptions. The
sets of indicators where there were not equal numbers of items in each indicator resulted in
more discrepant means (e.g., the personality facets and OCB). The indicators of work
withdrawal did produce a large discrepancy in means that is not so easily explained. Work
withdrawal indicator 1 produced a mean that was approximately seven points higher than the
other two indicators. Because of this difference the computation of the scale score was
double checked. It appears that the procedure for creating indicators resulted in a
coincidental combination of items which produced higher scores on work withdrawal
indicator 1.
Assessment of the measurement models was essentially conducted as confirmatory
factor analyses. The maximimi likelihood estimation procedure was used to estimate the
factor loadings for each of the manifest indicators on to its theoretical latent factor. The
maximum likelihood method was used becaiise it has been found to have "the desirable
asymptotic, or large-sample, properties of being unbiased, consistent, and efficient"
(Anderson & Gerbing, 1988, p. 413). Anderson and Gerbing also state that it has been found
to be robust against moderate violations of multivariate normality.
The measurement model analyses were conducted using a covariance matrix.
Joreskog and Sorbom (1989) suggest the use of a covariance matrix as a general rule in
structural equation modeling because of complications that can be encountered when using a
correlation matrix. They cite Cudeck (1989) who claimed that using a correlation matrix as
input can produce incorrect X" values. Schumacker and Lomax (1996) also recommend
using covariance matrices as input. They conclude that using correlation matrices as input
can lead to imprecise values for both the parameter estimates and error estimates. The main
68
Table 2. Means and Standard Deviations of the Manifest Indicators of the Latent Constructs
Indicator Mean Standard Deviation Neuroticism 1 (4) ^ Neuroticism 2 (4) 11.46 2.70 Neuroticism3 (4) II.66 2.63 Agreeableness 1 (4) 15.09 1.99 Agreeableness 2 (4) 15.44 2.14 Agreeableness 3 (4) 16.16 2.20 Conscientiousness 1 (4) 15.89 2.07 Conscientiousness 2 (4) 16.04 2.00 Conscientiousness 3 (4) 16.27 2.14 Impulsiveness 1 (3) 8.20 2.15 Impulsiveness 2(3) 8.87 1.87 Impulsiveness 3 (2) 7.12 1.20 Altruism 1 (3) 12.41 1.57 Altruism 2 (3) 12.12 1.37 Altruism 3 (2) 8.45 1.11 Dutifiilness 1 (3) 12.35 1.53 Dutiflilness 2 (3) 12.30 1.65 Dutiflzlness 3 (2) 8.10 1.50 Positive Affect 1 (4) 13.61 2.82 Positive Aifect2 (3) 10.11 2.19 Positive Affect 3 (3) 10.04 2.14 Negative Affect 1 (4) 7.17 2.59 Negative Affect 2(3) 5.19 1.80 Negative Affect 3 (3) 5.08 1.99 Job Satisfaction 1 (24) 49.08 11.95 Job Satisfaction 2 (24) 48.92 12.00 Job Satisfaction 3 (24) 49.64 12.86 Work Withdrawal 1 (7) 22.69 5.73 Work Withdrawal 2 (7) 15.54 5.88 Work Withdrawal 3 (7) 15.49 5.97 Organizational Citizenship Behavior 1 (6) 32.19 6.02 Organizational Citizenship Behavior 2 (6) 34.88 7.13 Organizational Citizenship Behavior 3 (5) 25.13 4.73
A^=313.
" The number of items in each subscale is presented in parentheses.
69
drawback to using a covariance matrix is the resulting difficulty of interpreting parameter
values that are dependent on the format of the measurement scale. The covariance matrix
used in the personality factor analyses is provided in Appendix J. The covariance matrix
used in the personality facet analyses is provided in Appendix K.
70
CHAPTERS: RESLXTS
The means, standard deviations, correlations, and reliabilities for all of the variables
of interest to the study are presented in Table 4. The coefficient alphas for the work
withdrawal and OCB scales were .81 and .83 respectively. These were somewhat higher than
expected given the intentional strategy of incorporating the breadth of behaviors that can
comprise a behavioral family. However, it suggests that die behaviors are internally
consistent and good representations of the constructs. The reliability estimate for the job
satisfaction measure was, as expected, very good (r = .94). The personality factor scales
reliability values were .79, .71, and .78 for neuroticism, agreeableness. and
conscientiousness, respectively. The personality facet scales produced the lowest reliability
estimates. The coefficient alphas were .63, .64, and .55 for impulsiveness, altruism, and
dutifulness, respectively. These lower values were expected based on the reliability estimates
reported by the scales' authors (Costa & McCrae, 1992) and the small number of items in
each scale. The values, however, are disappointing and may influence the confidence with
which conclusions can be made about the personality facets. Finally, the reliabilities for
positive and negative affect were .87 and .83, respectively.
Several of the correlations presented in the correlation table are worth noting. First is
the non-significant correlation (all significance tests in this study used an alpha level of .05)
between work withdrawal and OCB (r = .06). This provides support for concepmalizing
work withdrawal and OCB as separate constructs. If they were opposite ends of a single
construct, they would be expected to have a significant negative relationship.
Consistent with the proposed hypotheses, work withdrawal was significantly
correlated (both statistically and practically) with neuroticism, conscientiousness,
impulsiveness, dutifulness, job satisfaction, and negative affect. Inconsistent with the
hypotheses, however, is the fact that work withdrawal was also significantly correlated
(negatively) with agreeableness, altruism, and positive affect.
Table 4. Descriptive Statistics and Correlations Among Variables M SD 1 2 3 4 5 6 7
1. Full Time/I'art l ime 2. Sex -.01* 3. Marital Status .08* .08* 4. Age -.16 -.05* .14 5. Education -.20 -.04* -.19 -.08* 6. Salary -.28 -.27 -.11* .14 .41 7. Withdrawal (21) 49.16 14.87 -.18 .04* -.18 -.25 .18 .09* .81 8. Citizenship (17) 93.90 15.01 -.13 .05* .05* -.01* .06* .14 -.06* 9. Job Satisfaction (72) 147.64 35.23 .11* -.06* .02* .12 .03* .13 -.29
10. Neuroticism (12) 31.96 6.84 -.09* .15 -.05* -.21 -.09* -.20 .40 11. Agreeableness (12) 46.69 5.05 .06* .07* .01* .14 .04* -.01* -.37 12. Conscientiousness (12) 48.20 5.20 .08* .02* .11* .11* -.05* .03* -.49 13. Impulsiveness (8) 22.96 4.01 -.08* .12 -.11* -.12 -.01* -.12 .32 14. Altruism (8) 32.98 3.21 .11* -.01* .06* .01* .03* -.05* -.31 15. Dutiful (8) 32.75 3.43 -.01* .03* .07* .33 .07* .05* -.42 16. Positive Affect (10) 33.76 6.38 .03* -.01* .14 .15 .07* .10* -.30 17. Negative Affect (10) 17.44 5.62 -.02* .02* -.04* -.25 -.04* -.14 .27
Note: The number of items in each scale is presented in parentheses in the first column. Reliabilities appear in bold on the diagonal. * indicates non-significance (p > .05). N = 313.
Table 4. (Continued) 8 9 rn n n n r? n rr
1. Full i'ime/l'art Time 2. Sex 3. Marital Status 4. Age 5. Education 6. Salary 7. Withdrawal (21) 8. Citizenship (17) 9. Job Satisfaction (72)
10. Neuroticism (12) 11. Agreeableness (12) 12. Conscientiousness (12) 13. Impulsiveness (8) 14. Altruism (8) 15. Dutiful (8) 16. Positive Affect (10) .30 .42 -.37 .22 .37 -.17 .26 .32 .87 17. Negative Affect (10) .01* -.29 .54 -.34 -.30 .18 -.21 -.25 -.29 .83
.83
.11 .94 -.17 -.35 .79 .14 .32 -.36 .71 .28 .08* -.43 .27 .78
-.14 -.16 .49 -.25 -.27 .63 .26 .16 -.25 .65 .34 -.09* .64 .22 .12 -.39 .29 .62 -.22 .32 .55 .30 .42 -.37 .22 .37 -.17 .26 .32 .01* -.29 .54 -.34 -.30 .18 -.21 -.25
No(e: The number of items in each scale is presented in parentheses in the first column. Reliabilities appear in bold on the diagonal * indicates non-significance (/; > .05). N = 313.
7.3
Similar to work withdrawal, OCB was also significantly correlated with the
hypothesized variables. There were also, however, significant correlations with variables that
were hypothesized to not be related to OCB (e.g., neuroticism and impulsiveness). One
surprising relation was the low correlation between job satisfaction and OCB (r = .11).
Although it is statistically significant it is considerably lower than the correlations reported
in the literature which often range from .20 to .50 (see Fisher & Locke, 1992).
Correlations among the affect and personality variables ranged from . 17 to .54 in
absolute values. The personality variables most strongly correlated with positive affect was
neuroticism (r = -.37) and conscientiousness (r = .37). The personality variable least strongly
correlated with positive affect was impulsiveness (r = -.17). This is of special interest given
that impulsiveness is a subfactor of neuroticism. Therefore, they were expected to be have
similar relations.
The personality variable most strongly related to negative affect was neuroticism
{r = .54). Interestingly, the personality variable least strongly correlated wdth negative affect
was impulsiveness. This correlation suggests that neuroticism and affect are similar
constructs and may not be as distinct as has been implied in the literature.
Series I.
The results from analyzing the models in series 1 (the models using personality factors) will
be presented first. Factor loadings and error estimates for the indicators in the measurement
model are presented in Table 5. Because a covariance matrix was used for the analyses, the
unstandardized estimates are difficult to interpret due to their being dependent on non-
standardized scale scores. Both standardized and unstandardized estimates are provided to
document the non-standardized values. The standardized estimates are interpreted as the
correlations between the manifest indicators and their latent variables. Squaring the
standardized estimate provides the amount of variance accounted for by the indicator. The
standardized error estimate is then equal to one minus the variance accounted
74
Table 5. Factor Loadings for Manifest Variables in Measurement Model For Series I
Unstandardized Unstandardized Standardized Standardized _ , r ,• ML Estimate Error Estimate ML Estimate Error Estimate Constructs and Indicators
Work Withdrawal WWI 1.00 17.31 .69 .53 WW2 1.31 7.98 .88 .23 WW3 1.19 13.73 .78 .39
Organizational Citizenship Behavior OCBl 1.00 8.35 .88 .23 0CB2 .96 25.26 .71 .50 0CB3 .70 8.63 .78 .39
Neuroticism Neurotl 1.00 3.00 .78 .40 NeurotZ .94 3.25 .74 .45 Neurotj .84 3.63 .69 .53
Agreeabieness Agree 1 1.00 1.87 .56 .69 Agree2 1.86 3.24 .69 .52 Agree3 1.80 2.11 .75 .43
Conscientiousness Consc I 1.00 2.09 .72 .49 Consc2 1.00 1.80 .74 .45 Consc3 l . l l 1.84 .77 .40
Positive Affect PosAffl 1.00 2.14 .85 .27 PosAfE2 .75 1.59 .82 .33 PosAfD .73 1.53 .82 .33
Negative Affect NegAffl 1.00 2.08 .83 .31 NegAffZ .66 1.25 .78 .39 NegAff3 .75 1.35 .81 .34
Job Satisfaction JobSatI 1.00 19.89 .93 . 1 4 JobSat2 1.03 14.06 .95 . 1 0 JobSat3 1.07 23.40 .93 . 1 4
Note: N = 313 All values significant p < .05
75
for. That is, the standardized error estimate is the amount of variance in the manifest
indicator not accounted for by the latent factor.
The factor loadings and estimates presented are consistent with other indications of
the psychometric quality of the measures. The best factor loadings are those produced by the
indicators of job satisfaction. All three indicators produced standardized estimates above .90.
The worst factor loadings in the model are associated with the personality factor indicators.
Their standardized values ranged from .56 to .78. All of the factor loadings shown in Table 5
were significant at thep < .05 level
Goodness-of-fit indices for the models tested in series 1 are presented in Table 6. The
measurement model produced satisfactory indices of fit. The Idf ratio was 2.24 which is
considered acceptable (Schumacker & Lomax, 1996). The GFI of .88 is also considered to
be acceptable. Normal convention suggests GFI values of .90 or above indicate a good fit.
Given that .90 is a convention and not a strict criterion, the value of .88 was determined to be
acceptable. The AGFI was .84 which was also considered to be adequate. Authors have
suggested that AGFI values of .80 or better are usually an indication of acceptable fit
(Pedhauzer & Schmelkin, 1991). The standardized RMSR value was .07. A value of .05 is
generally considered acceptable (Schumacker & Lomax. 1996). These indices are consistent
in the fact that they all are considered satisfactory.
As was stated previously, the measurement model was estimated first to allow for the
separate evaluation of the measurement and structural models. After estimating the
measurement model, the covariance in the measurement models were, in effect, held stable.
This allows for the testing of the structural model as a regression procedure, only using latent
variables. Therefore, the path coefficients in the structural model can be interpreted as
similar to the standardized regression coefficients for the latent variables.
The next step in the analysis consisted of using the unstandardized factor
loadings produced in the measurement model to estimate the latent structural models. The
16
first model tested was the null model. It included no causal paths but allowed for each of the
independent variables to be correlated (see Figure 10). Fit indices for this model are
provided in Table 6. The for Model 1 was 2.42. The main purpose of this model
is to provide a baseline to which fit indices firom subsequent models can be compared. As
expected. Model I was the worst-fitting model tested. It produced the lowest GFI and AGFI
of all models tested (.85 and .84 respectively). It also produced the largest RMSR value(. 13)
of all models tested. Allowing the endogenous variables to correlate is supported by the
relatively high correlations among them. Only conscientiousness and job satisfaction were
correlated below .25 in absolute value.
Table 6. Goodness-of-Fit Indices for Measurement and Structural Models in Series 1.
Model f df -///df GH AGFI RMSR
Measurement 501.13 224 224 .88 .84 .07 Model 1 670.26 277 2.42 .85 .84 . 1 3 Model 2 646.26 275 2.35 .86 .84 . 1 1 Model 3 613.82 273 2.25 .86 .85 .09 Model 4 547.75 269 2.03 .87 .86 .08
In Model 2. causal paths firom job satisfaction to work withdrawal and OCB were
estimated (see Figtire 11). This produced a statistically significant drop in fi'om 670.26 to
646.26, Ax" (2, .'V= 313) = 24,;? < .05. Other fit indices also improved but would still not be
considered to demonstrate a good fit. The ratio for Model 2 was 2.35. The GFI and
AGFI were .86 and .84, respectively, and the RMSR was .11. The path coefficients were -.30
(p < .05) for job satisfaction to work withdrawal and .05 for job satisfaction to OCB {p >
.05). The lack of significance for the path between job satisfaction and OCB was unexpected
and contrary to the hypothesized relationship. The squared multiple correlations for the
structural equations were .09 and 0.00 for work withdrawal and OCB, respectively. In other
words, job satisfaction accounts for nine percent of the variance in work withdrawal and none
of the variance in OCB.
Neurolicism
Fnci(»r
Conscientiousness l-Hclor
Agrecableness
l-'acl(ir
Negative
Affect
Job
Satisfuctiiin
Positive
Affect
I'igure 10. Model 1: 'i'he Null Model I'or Series 1.
Work
Withdrawal
Organi/atioiml Cili/enship
Behavior
Ncumlicisin
I actor
Conscicnlioiisnuss
laclor
Work
Wilhdniwal
Apreenbluness
I'aclor
Negative
Alfccl
.27 Organizationul
Citizunsliip nelmvior
Job
Satisruction 05 (.86)
R- = 0.00
Positive
A fleet
Figure 11. Model 2 for Series 1: Showing the causal relationship of job satisfaction on work withdrawal and OCB. Completely standardized path coefllcients are shown with t-values in parentheses.
•45
/?-= .14
Neuroticism Factor
-5J
.64
.37
- .37 Agreeableness
Faclor -42
.25(3.71)
.10 - .43
-.21(3.37)
Negative
Affect
- .32 .27 -.10(1.44)
- .35
.32 (4.42) 46
Positive Affect
Conscientiousness V. Factor >
= .08
Figure 12. Model 3 lor Series 1: Positive and negative affect causal paths. Completely standardized path coefficients are shown with t-values in parentheses.
80
The additional paths freed in Model 3 were from positive affect to OCB and from
negative affect to work withdrawal (see Figure 12). The path from job satisfaction to OCB
continued to be not statistically significant. From Model 2 to Model 3, there was an expected
drop in from 646.26 to 613.82. The change was statistically significant A^" (2. :V = 313) =
32.44, p < .05. The fit indices also improved as expected. The x^df ratio for Model 3 was
2.25. The GFI did not change from Model 2 and remained at .86. The AGFI did. however,
move up to .85 and the RMSR was reduced to .09 from .11. There was also an expected
improvement in the squared multiple correlations. Variance accounted for increased by .05
for work withdrawal from Model 2 (/?" = . 14) and by .08 for OCB {FT = .08).
The final model tested in series 1 freed the following paths: I) neuroticism to work
withdrawal, 2) conscientiousness to both work withdrawal and OCB, and 3) agreeableness to
OCB. Figure 13 shows the model with the resulting path coefficients. The paths from
conscientiousness and job satisfaction to work withdrawal and the path from positive affect
to OCB were statistically significant. The remaining paths were not.
The lack of significance for neuroticism and agreeableness was inconsistent with the
hypotheses. One possible explanation may be found by examining the correlations between
neuroticism, agreeableness, and affect. Neuroticism is correlated .64 with negative affect.
Neither of them resulted in a significant effect on work withdrawal in Model 4. Negative
affect, however did have a significant effect on work withdrawal in Model 3. The high
correlation between the two variables and the influence of negative affect changing to non
significant suggests that the two predictors may be competing for criterion variance. As a
result, neither of them reach significance. A similar situation may exist for agreeableness and
positive affect. They were correlated .27. While positive affect did produce a significant
effect in Model 4. it was reduced from .32 in Model 3 to .28. This too, could suggest that
agreeableness is unable to predict beyond what is afready accounted for by positive affect.
- .45
Neurolicism
Factor
-53
.19(1.79) Conscientiousness
V Factor y -.46(5.91) Work
Withdrawal .64
.37 -.08(.97)
Agreeableness
Factor - .37
- .42
-.19(3.09)
-43 .07(.81)
Negative Affect
37
- .32 Organizational Citizenship
Behavior
.-.09(1.20)
Job
Satisfaction - 35
.28(3.23)
Positive Affect
R- = M
Organizational \ d 2 _ no Citizenship ' "
Behavior
00
l-igure 13. Model 4 for Series 1; Personality l-actors. Completely standardized path coefllcients are shown with t-values in pnrcnlncscs.
82
This model produced the largest drop in for the four models tested in the
personality factor series. Once again, the model improvement produced a statistically
significant change in x"5 313) = 66.07, p < .05. The x," for this model was
547.75 with 269 degrees of freedom. This resulted in a x'^df ratio of 2.03. The GFI was .87
and the AGFI was .86. The RMSR was .08. The squared multiple correlations for the
structural equations in this model were .37 and .09 for work withdrawal and OCB.
respectively.
The fit indices indicate a good fit for the final model of this series. Although the x~
value was statistically significant for all of the models tested, this is to be expected given the
size of the sample (Schumacker & Lomax, 1996). The important result of testing the series
of models is the incremental improvements realized from one model to the next. The
improvements were consistent with the proposed relationships and established the additional
contributions of each of the sets of variables.
The second series of models tested followed the same procedures as the first. The
only difference was in the measurement of the personality variables included in the models.
Instead of using the personality factors of neuroticism, agreeableness. and conscientiousness,
the personality facets of impulsiveness, altruism, and dutifiilness were used in their places.
The goodness-of-fit indices for the models tested in series two are presented in Table 7.
The measurement model for this series produced a satisfactory level of tit. The yj
with 224 degrees of freedom was 462.08 which resulted in a x"/ df ratio of 2.06. The GFI
was .89 and the AGFI was .86. The standardized RMSR was .06.
Table 7. Goodness-of-Fit Indices for Measurement and Structural Models in Series 2.
Model X" df X"/df GFI AGFI RMSR
Measurement 462.08 224 2.06 .89 .86 .06 Model 1 620.43 277 2.24 .87 .86 . 1 2 Model 2 596.51 275 2.T7 .87 .86 . 1 0 Model 3 566.78 273 2.08 .87 .86 .09 Model 4 495.53 269 1.84 .88 .87 .07
83
The factor loadings produced for the measurement model from series two are
presented in Table 8. All of the factor loadings for the model were significant at the p < .05
level. Again, the best factor loadings are those produced by the indicators of job satisfaction.
All three indicators produced standardized estimates above .90. The weakest factor loadings
in the model are associated with the personality indicators. The standardized values for the
personality facets ranged from .46 to .84.
Similar to the first series of models, the first structural model tested was a model with
no causal paths (see Figure 14). Only the correlations among the independent variables were
allowed to be estimated. The results produced a x" of 620.43 with 277 degrees of freedom.
This provided a df ratio of 2.24, a GFI of .87, an AGFI of .86, and a RMSR of. 12. This
was, as expected, the worst-fitting model of the series.
The second model included the causal paths from job satisfaction to work withdrawal
and OCB. It expectedly produced a statistically significant decrease in the chi-square (Ax"
(2, = 313) = 23.92, p < .05) from Model I. The chi-square value of 596.51 resulted in a x"'
df ratio of 2.17. The GFI and AGFI did not change from Model I and remained at .87 and
.86, respectively. The RMSR did improve from .12 to .10. The squared multiple correlation
for the structural equations were the same as reported in the first series for this model
{FT = .09 for work withdrawal and R" = 0.00 for OCB).
Structural model estimates for Model 2 are presented in Figure 15. The path
coefficient for job satisfaction to work withdrawal was -.29 (p < .05) and was statistically
significant as expected. The path coefficient for job satisfaction to OCB was .05 and was not
statistically significant. As was stated in the first series of models, this was an unexpected
result. Job satisfaction was hypothesized to be significantly related to OCB.
In addition to freeing the job satisfaction paths shown in Model 2, Model 3 also
estimated the causal paths from positive affect to OCB and from negative affect to work
withdrawal.
84
Table 8. Factor Loadings for Manifest Variables in Model Series 2
Unstandardized Unstandardized Standardized Standardized „ ., ML Estimate Error Estimate ML Estimate Error Estimate Constructs and Indicators
Work Withdrawal WWl 1.00 17.82 .68 .54 WW2 1.36 6.86 .90 .20 WW3 1 . 1 9 14.27 .77 .40
Organizational Citizenship Behavior OCBl I.GO 8.04 .88 .22 0CB2 .95 25.40 . 7 1 .50 0CB3 .70 8.78 .78 .39
Impulsiveness Imp I 1.00 1.39 .62 .62 Imp2 1.68 1.04 .84 .30 Imp3 .79 1.98 .46 .78
Altruism A l t l 1.00 1.48 .63 .60 AIt2 .96 .99 .69 .52 Alt3 .74 .70 .66 ^57
Dutifulness Dutl 1.00 1.50 .60 .64 Dut2 .93 1.99 .52 .73 Dut3 .88 1.60 .54 . 7 1
Positive Affect PosAffl 1.00 1.97 .87 .25 PosAfD .73 1.64 .81 .34 PosAfD . 7 1 1.60 .81 .35
Negative Affect NegAffl 1.00 2.07 .83 . 3 1 NegAff2 .65 1.27 .78 .39 NegAfD .75 1.33 .81 .34
Job Satisfaction JobSatl 1.00 20.00 .93 . 1 4 JobSat2 1.03 13.93 .95 . 1 0 JobSatS 1.07 23.44 .93 . 1 4
N = 3I3 All values significant p < .05
Impulsiveness
l-acel
- .30
Dutilulness
Facet
.25
.55
- .37 - 1 8
- .27 - .23
Negative
Affect
46
- .32 .35
Job
Satisfaction - .34
46
Positive
Affect
I-'igure 14. Model 1: Tiie Null Model for Scries 2.
Work
Withdrawal
Organizational Citizenship
Behavior
•23
Impulsiveness
I'acet
- 30
- .10
Diitifulness
Facet
Work
Withdrawal 25
-37 Altruism
Facet - 1 8 -.29(4.90
-27
Negative
AfTecl
-32 Organizational Citizenship
Behavior
.05(.88) .35
Job
Satisfaction - .34
Posilive
Affect
= .09
Organizational \ ni _ n nn Citizenship ^ ^ ~ "
Behavior
00 0\
Figure 15. Model 2 Tor Seric.s 2: Showing (he causal relationship of job standardized path coefficients are shown with t-values in parentheses.
satisfaction on work withdrawal atid OCB. Conipletely
87
As expected. Model 3 produced a decrease in chi-square that was statistically significant. Ax"
(2. 313) = 29.73.< .05. This better-fitting model produced a x"/ df ratio of 2.08. The
GFI and AGFI values did not change from Model 2 and were .87 and .86, respectively. The
RMSR did decrease to a value of .09 from . 10. The improvement in the variance accounted
for is very similar to Model 3 in the first series of models (R^ = .13 and R~ = .08 for work
withdrawal and OCB, respectively).
The paths and path coefficients for Model 3 are shown in Figtire 16. Like in Model 2.
the relationship between job satisfaction and work withdrawal was significant while the
relationship between job satisfaction and OCB was not. However, comparing the coefficient
values associated with these relationships in Model 2 and Model 3 reveals that the job
satisfaction - work withdrawal coefficient decreased from -.29 to -.22 and the job satisfaction
- OCB coefficient changed signs and decreased from .05 to -.10. The path coefficient from
negative affect to work withdrawal was .22 (p < .05). The path coefficient from positive
affect to OCB was .32 (p < .05). The affect results were consistent with the hypothesis
suggesting that individuals' affect level will influence the frequency of non-workrole
behaviors.
Finally. Model 4. which was the model hypothesized to have the best fit was tested.
It freed the paths from impulsiveness to work withdrawal, dutifulness to both work
withdrawal and OCB, and from altruism to OCB. Freeing these paths produced an expected
change in chi-square that was statistically significant (A^" (4, 313) = 71.25, p < .05).
Also, the resulting fit indices were the best of all models tested in either series. The x~/ df
ratio dropped below 2.0 to 1.84. The GFI and AGFI increased from .87 to .88 and from .86
to .87, respectively, and the RMSR dropped to from .09 to .07. The amount of criterion
variance accounted for in this model is the highest of all eight models tested. The squared
multiple correlations were.43 and. 16 for work withdrawal and OCB, respectively.
-23
Impulsiveness
lacel
- .30
Dulifuiness
i'acet
-.10
Work
Withdrawal 25 .55
.22(3.31) Altruiim
Facet - 1 8
- .27 -.22(3.49)
Negative
Affect
Organizational Citizenship
^ Behavior
- .32 -.10(1.43) .35
Job
Satisfaction - .34
.32(4.41)
Positive
Affect
/?- = .13
Organizational \ p2 _ (\o Citizenship ^
Behavior
00 00
Figure 16. Model 3 for Series 2: Positive and negative affect causal paths. Completely standardized path coefficients are shown with t-values in parentheses.
89
The path coefficients for Model 4 are shown in Figtire 17. All of the paths except the
path from negative affect to work withdrawal and the path from job satisfaction to OCB were
statistically significant. Therefore, the results generally support the hypotheses. One
exception is the relationship between job satisfaction and OCB. This relationship was
consistently disappointing across all of the models tested. The path from dutifulness to OCB
was significant, but the coefficient was negative instead of positive as hypothesized. Another
exception was the path from negative affect to work withdrawal; it was not significant in this
model. This path was, however, significant in Model 3. This would suggest that negative
affect is sharing variance with one or more of the personality facets entered in Model 4.
Examination of the correlations between negative affect and the personality facets reveals a
correlation of .25 between negative affect and impulsiveness. This likely indicates that work
withdrawal variance predicted by negative affect in Model 3 is now being predicted by
impulsiveness in Model 4.
Similar to the first series, the incremental improvements in each successive model in
this series indicate support for the hypothesized relationships. Overall, the results of the
second series of models were superior to the results of the first series. The path coefficients
were of greater absolute value, and more paths were as statistically significant. The total
variance accounted for in the criterion variables was also higher than in the first series of
models. The first series of models produced final variance accounted for statistics of R~= .37
and R~ = .09 for work withdrawal and OCB, respectively. The second series of models
resulted in R' values of .43 and .16 for work withdrawal and OCB, respectively.
-.23
Impulsiveness
I'acet
.17(2.50) - .30
- 1 0 -.54(6.41) Work
Withdrawal
.25 -.05(.74) .55
Altruism
Facet -.19(3.04) -37 - 1 8
-.27
-.22(1.97) Negative
Affect 46
.33(3.29)
Organizational Citizenship
Behavior
-32 -.13(1.84) .35
Job
Satisfaction -34
.31(3.64) 46
Positive
Affeit
/?• = .43
OrganizationalX = i a Citizenship
Behavior
I'igure 17. Model 4 for Series 2; Personality lacets Completely standardized path coeincients are shown with t-values in parentheses.
so O
91
CHAPTER 6: DISCUSSION
This study has contributed to the areas of personality, job satisfaction, organizational
citizenship, and organizational withdrawal research by providing a unified, multivariate
examination of the affects of job satisfaction, positive and negative affect, and criterion-
related measures of personality on broadly-measured constructs of employee behavior
labeled here as non-workrole behaviors. Previous studies of personality and organizational
outcomes such as employee performance, turnover, and/or absenteeism have suffered from
various methodological problems. This study has attempted to avoid the errors of previous
studies by using generally accepted measures of personality and job satisfaction as
independent variables. It also evaluated a broader and more appropriate measure of the
dependent variables than previous studies have used.
The main hypotheses for this study centered around the contributions of personality to
the prediction of work withdrawal and OCB. Past studies examining the effects of
personality on employee behavior have used conceptualizations of personality other than
what is currently accepted in the five-factor theory of personality. By conceptualizing
personality in terms of the five-factor model, this study brings non-workrole behavior into
the same area of focus as has recently been the case with personality and traditional work
performance (e.g.. Barrick & Mount, 1991; Schneider & Hough, 1995; Tett, Jackson. &
Rothstein. 1991). This study has also attempted to bring together what have typically been
separate lines of research. Many studies have examined subsets of what is presented here,
but none have addressed them simultaneously in a single model. That is, studies have
examined job satisfaction and absenteeism or job satisfaction and OCB. Studies have also
examined the relationship of personality to these constructs. There have not. however, been
smdies that have addressed them all in a single framework.
The general findings from the final models tested in both series suggest that
personality, affect, and job satisfaction have an important role in the prediction of non-
92
workrole behaviors. Examination of multiple fit indices provides a consistent indication of
adequate fit of the hypothesized models. More importantly, when the models within each
series are compared, the residts are consistent with the hypotheses which expected an
increase in model fit statistics as each of the theoretical constructs was added to the models.
The chi-square values decreased by statistically significant amounts as the models were built
from job satisfaction only, to job satisfaction and affective state, and finally to job
satisfaction, affective state, and personality.
Within each of the models, the patterns of path coefficients provide more specific
information about the relationships between the variables modeled in each series. Therefore,
the individual relationships of each of the variables represented in the models are addressed
below.
Job Satisfaction
As expected, job satisfaction did have a negative relationship with work withdrawal.
The strength of the relationship, however, decreased somewhat as affect and personality
variables were added to the model. This indicates that job satisfaction shares variance to
some extent with affective state and the various forms of personality. The substantial
correlations between job satisfaction and several of the affect and personality variables also
supports this finding. Neuroticism, agreeableness, positive affect, and negative affect
produced correlations with job satisfaction that ranged from .29 to .42 in absolute value (see
Table 4).
The correlation between job satisfaction and positive affect was .42. The correlation
between job satisfaction and negative affect was -.29. When affect was added to the model,
the path coefficient for the path from job satisfaction to work withdrawal decreased from -.30
to -.21 in the first series of models and from -.29 to -.22 in the second series of models.
However, both job satisfaction and positive and negative affect still produced significant
structural paths to work withdrawal. This suggests that, although related, job satisfaction and
93
affect are each making unique contributions to the prediction of work withdrawal. This
finding also provides some support for Brief and Roberson's (1989) claim that traditional
measures of job satisfaction such as the JDl are focused on cognitive considerations of
satisfaction and do not include an emotional or affective element that, according to the data
presented here, also contributes to the determination of employees' non-workrole behaviors.
The next components added to the model were the personality factors. The addition
of the personality factors in series 1 resulted in conscientiousness showing a strong
relationship to work withdrawal behaviors (-.46). While adding personality factors resulted
in a large increase in the variance accounted for in work withdrawal (A/?" = .23), the path
coefficient between job satisfaction and work withdrawal remained approximately the same
(the value changed from -.21 to -.19). This suggests that personality, specifically
conscientiousness, plays an important role in the prediction of work withdrawal behavior.
These results have important implications for fiiture research in job satisfaction and
organizational withdrawal. The most important implication is the importance of individual
differences in organizational withdrawal theory. Individual differences have been
conspicuously absent from the most popular theories of organizational withdrawal. For
example. Hulin's model of organizational withdrawal, which was presented in Figxore I.
considers many intemal and external antecedents of withdrawal such as individual skills and
economic conditions. It does not, however, include any consideration of individuals'
personality. The path coefficient of -.46 from conscientiousness to work withdrawal is strong
evidence of the part that personality plays in the determination of work withdrawal behaviors.
Job satisfaction was also hypothesized to be related to OCB. The results however did
not support this hypothesis. The path from job satisfaction to OCB in Model 2 in both series
of models was only .05. Similarly, the zero-order correlation for general job satisfaction and
OCB in this study was only Al {p < .05). Why the relation is so weak is unclear. One
possible explanation stems from the composition of the job satisfaction scale. The measure
94
of general job satisfaction used in this study was a combination of the five subscales within
the JDI. The low correlation may be the result of conflicting relationships between OCB and
one or more of the individual JDI subscales.
To investigate this relationship, an additional analysis was conducted examining the
correlations among OCB and each of the five dimensions within the measure of general job
satisfaction. This analysis revealed that there was only one statistically significant
relationship among OCB and the five dimensions. The significant correlation was between
OCB and satisfaction with the work itself (r = .21). All four of the other correlations were
non-significant. Speculation regarding why OCB was only related to satisfaction with the
work itself might suggest that it is most salient of the five elements of job satisfaction when it
comes to determining the influence of job attitudes on engaging in positive non-workrole
behaviors.
A similar analysis was conducted using work withdrawal in order to compare its
relation to the five facets of job satisfaction. The results were quite different. Four of the
five correlations were statistically significant. The non-significant correlation was between
work withdrawal and satisfaction with pay and benefits (r = -. 10). The other correlations
were -.21, -.27, -.22, and -.19 for work, promotions, supervisor, and co-workers, respectively.
These results provide additional support for the distinction between work withdrawal and
OCB.
OCB's low correlations across all of the variables in the study suggest further
investigation of the OCB measure may be necessary. Examining the correlations of OCB
with the other variables in the study reveals that OCB is not correlated higher than .30 with
any of the variables. OCB is most highly correlated with positive affect (r = .30). The
variables with the lowest correlation with OCB are age and negative affect (r = -.01 and r =
.01, respectively). Possibly future research could examine the factor structure of this
95
particular measure of OCB to determine if multiple factors could be contributing to the lack
of expected correlations.
Affective State
Affective state also played a significant role in the prediction of both work
withdrawal and OCB. In this study, affective state was operationalized as the extent to which
participants reported experiencing various emotions or moods during the week prior to
completing the questionnaire.
The significant paths from positive affect to OCB in both series of models suggest
that a person's mood influences his or her performance of OCB. It is not difficult to imagine
the situation in which an employee has the opportunity to perform a voluntary positive
behavior at work and how the decision to carry out the behavior may be influenced by his or
her recent mood. For example, staying late for a few minutes to help a co-worker may be
more likely for a person in a good mood than a person in a bad mood.
While negative affect did significantly influence work withdrawal in Model 3 of both
series, the path from negative affect to work withdrawal was non-significant in the final
model (Model 4) of both series. In series I, this result could be explained by negative affect's
high correlation with neuroticism =.64). It appears that the variance in work withdrawal
previously predicted by negative affect in model 3 is now being accounted for by neuroticism
in Model 4. In series 2. however, the correlation between negative affect and impulsiveness
(a facet of neuroticism) is relatively quite small {(p =.25). Therefore, this explanation does
not hold for the negative affect - work withdrawal path being non-significant in series 2
The relations or lack of relations between positive affect, negative affect, job
satisfaction, work withdrawal, and OCB in this study are particularly interesting. It appears
that OCB is more strongly influenced by affective state (positive affect) than by a general
attitude (job satisfaction). In contrast the opposite is true for work withdrawal. It is more
96
strongly influenced by a general attitude (job satisfaction) than by affective state (negative
affect).
Although previous research has not compared OCB and work withdrawal together,
the findings regarding OCB and affect are consistent with previous studies of affect and OCB
(e.g., George, 1991; Organ & Konovsky. 1989; Organ & Lingl. 1995). A possible direction
for future research is the flulher investigation of these relationships to corroborate the
findings of this study.
Personality
In the first series of models, the strongest relationship produced was the effect of the
personality measure of conscientiousness on work withdrawal. The path coefficient of -.46
suggests the predictable conclusion that people who are scrupulous and reliable are less likely
to engage in behaviors that remove them from their regular work tasks than people who are
unreliable or careless. The significant relationship demonstrated between conscientiousness
and work withdrawal is consistent with other studies that have reported relationships for
conscientiousness and more traditional performance measures (e.g., Barrick & Mount. 1991:
Hogan, 1991).
The other two personality factors did not play a significant role in the prediction of
either positive or negative non-workrole behaviors. The paths from agreeableness to OCB
and neuroticism to work withdrawal were not statistically significant. These findings were
inconsistent with the hypotheses. In addition, they go against intuitive expectations based on
the definitions of the constructs.
For example, agreeableness is defined as being helpful and altruistic. OCB is made
up of behaviors that are generally helpful to the organization. Given this similarity, a logical
conclusion suggests that they should be related. It is not immediately apparent why the
hypothesis was not supported. It is possible that the result is a specific anomaly. However.
97
given the absence of previous research investigating this specific relationship between
agreeableness and OCB, it is difficult to make any conclusions.
Similarly, neuroticism did not fulfill its theoretical expectations. Neuroticism is
described as being prone to worry, anger, impulsiveness, and depression. It was
hypothesized that these tendencies would result in more frequent work withdrawal behaviors.
This theory, however, appears to be incorrect. Neuroticism was not significantly related to
work withdrawal.
Conscientiousness was also hypothesized to cause OCB. The data did not, however,
support thi s h y p o t h e s i s . T h e p a t h c o e f f i c i e n t w a s . 0 7 a n d w a s n o t s t a t i s t i c a l l y s i g n i f i c a n t ( p >
.05). This was somewhat surprising given the success reported in the literature of using
conscientiousness as a predictor of traditional job performance (e.g. Barrick & Mount. 1991).
Even though the path coefficient in the final model was not significant, the zero-order
correlation for conscientiousness and OCB was .28 (p < .05). The non-significant path
coefficient suggests that another variable (or other variables) may have accounted for the
variance that contributed to the zero-order correlation between conscientiousness and OCB.
Based on the relationships within the model, it appears that positive affect may be that
variable. Positive affect would therefore reduce the variance in OCB available to be
predicted by conscientiousness. As was noted previously, positive affect does have a
significant causal influence on OCB in Models 3 and 4 in both series of models. It is also
correlated with conscientiousness (r = .37, ^ .44).
Comparison of the Two Series of Models
Important differences are apparent in comparing the use of personality factors and
personality facets. As expected, the models were identical and had similar results until
Model 4 where the personality variables were added. Overall, the results suggest that the
personality facets are more effective than the personality factors in predicting non-workrole
behaviors. Six of the eight paths tested in the second series produced statistically significant
98
values (compared to only three in the first series), and the goodness of fit indices were
slightly better in series 2 than they were in series 1.
Impulsiveness, which is a specific component of neuroticism, produced a path
coefficient of .17 to the criterion of work withdrawal {p <.05). Unlike neuroticism fi-om
series one, the path for impulsiveness to work withdrawal is statistically significant.
Impulsiveness is described as the inability to resist urges or cravings (Costa & McCrae.
1992). The hypothesis for the relationship between impulsiveness and work withdrawal was
based on the theory that employees engage in withdrawal behaviors to cope or adapt to
negative job attitudes and/or cognitions regarding their jobs. These negative feelings create
an urge or craving to relieve them. Therefore, employees scoring highly on the
impulsiveness scale have a greater inability to control these urges and, as a result, engage in
more work withdrawal behaviors than employees with low impulsiveness scores. Employees
with low impulsiveness scores are better able control the temptations of withdrawal behaviors
such as coming back late after lunch or calling in sick. As a result, they engage in fewer
withdrawal behaviors.
The result of impulsiveness being significantly related to work withdrawal while
neuroticism was not is also support for die use of criterion-related conceptualizations of
personality. These results suggest that the general personality construct of neuroticism
includes content that is not criterion-related. For example, the neuroticism factor is
composed of facets such as self-consciousness and vulnerability that are not expected to be
related to work withdrawal. Given these results in which neuroticism is not a significant
predictor of work withdrawal, possibly the other elements of neuroticism such as anxiety and
anger are not related to work withdrawal as was originally hypothesized. This would explain
why neuroticism was not a significant predictor of work withdrawal and impulsiveness was.
Impiilsiveness was more closely related to the criterion variable than the other elements of
neuroticism. When combined with the other elements of neuroticism, the covariance
99
between work withdrawal and impulsiveness is eliminated by the lack of covariance between
work withdrawal and the other elements of neuroticism.
Next, the predictive performance of conscientiousness and dutiflilness is compared.
Conscientiousness was, as expected, significantly related to work withdrawal. Dutiflilness.
however, was a significant predictor of work withdrawal but also produced an unexpected
relationship with OCB. The coefficient for the path from dutiflilness to work withdrawal was
larger than the same path using conscientiousness (-.46 vs. -.54). This was consistent with
the hypotheses and provides support for the importance of using criterion-related measures of
personality.
Contrary to the hypothesis, dutifulness resulted in a negative relationship with OCB.
The negative path coefficient (-.22) between dutifulness and OCB is inconsistent with the
hypothesis and the logic that led to the hypothesis. High scorers on the dutifulness scale are
said to "adhere strictly to their ethical principles and scrupulously fulfill their moral
obligations" (Costa & McCrae, 1992, p. 18). Given that, by definition, OCB consists of
behaviors that are not required or obligatory, possibly those employees high in dutifulness do
not see OCB as part of their duty. Therefore, performing OCB is less likely to be performed
by those employees high in dutifulness. This explanation, however, would also suggest that
those employees who are low in dutifulness would be more likely to engage in OCB than
employees who are high in dutiflilness. Being low in dutifulness is described by Costa and
McCrae (1992) as being more casual about principles and obligations. Possibly this results in
a greater opportunity to participate in spontaneous non-workrole behavior. For example, if
an employee does not feel an obligation to complete his or her assigned work tasks, he or she
may spend the time engaging in positive non-workrole behavior. This is, however, purely
conjecture.
The third personality facet, altruism, performed differently than the general factor
agreeableness by being significantly related to OCB while agreeableness was not. The
100
significant path from altruism to OCB does support the theoretical, and expected, relationship
between being altruistic and performing non-workrole behaviors that help the employing
organization. The conclusion is fairly simple. People who tend to be helpfiil (i.e.. are
altruistic) will be helpful at work (i.e., engage in OCB).
Similar to neuroticism and impulsiveness, finding a relationship between altruism and
OCB and not a finding a relationship between agreeableness and OCB is likely connected to
the issue of using criterion-related personality constructs. While agreeableness was
theoretically linked to OCB, the results of this study suggest that its conceptualization was
too broad to demonstrate how it was related to OCB. The definition of agreeableness
includes concepts described by Costa and McCrae (1992) as trustworthiness,
straightforwardness, altruism, compliance, modesty, and tender-mindedness. Of these
subcategories of agreeableness, altruism was hypothesized as being most closely related to
the OCB criterion. As was mentioned with neuroticism and impulsiveness previously, the
covariance between altruism and OCB is not detectable when measured with the non-
criterion related elements of agreeableness.
Limitations
Although the findings of this study are important, there are a few limitations that
should be acknowledged. The first limitation is direcdy tied to one of the greatest benefits of
the smdy. It is the result of using the Big Five personality framework and the NEO FFI and
NEO PI-R facet scales as personality measures. This was one of the greatest strengths of the
study because it provided a link between the study of employees' non-workrole behavior and
mainstream five-factor personality research. The limitation comes from the low reliability
estimates produced by the personality scales. As was noted in the results section, the
personality scales produced the lowest factor loadings in the measurement models. This may
have contributed to the resulting quality of the measurement models. While the fit indices
were adequate, they ideally would have been better. The facet scales in particular are suspect
101
because of the small number of items in each scale and the restilting low reliabilities. The
alphas for the facet scales were .63, .64. and .55 for impulsiveness, altruism, and dutifulness.
respectively.
Examination of the modification indices reveals some tendency for cross loading on
some of the personality facet indicators. For example, the impiUsiveness indicator 2 shows a
completely standardized expected change for lambda x of -.28 if allowed to load on
dutifiibiess. Interestingly, the impulsiveness indicator 3 shows a completely standardized
expected change for lambda x of .28 if allowed to load on dutifiilness. Note the change from
negative to positive. Similarly, two of the manifest indicators of dutifulness show
completely standardized expected changes of .29 and -.22 if allowed to load on altruism.
These results suggest that the personality facet measures are not pure measures of
impulsiveness, dutifuhiess, and altruism. As a result, confidence in the personality facet
relationships demonstrated in this study are somewhat diminished. It does, however, provide
a good launching point for future research in this area. By using the personality facet scales
as a core, new scales could be developed that are similar in concept but more
psychometrically sound than the eight-item scales used in this study.
A second limitation stems from the use of self-report data. There are very few
options other than self-report data for the measurement of job satisfaction and affectivit}'.
Personality and non-workrole behaviors could, however, be assessed by someone other than
the individual performing the behaviors. For example, co-workers or supervisors could rate
an employee's non-workrole behaviors, and a fnend or family-member could rate the
individual's personality. These measures were logistically not possible in this particular
smdy. but including them in future research would improve the reliability and validity of
those variables and strengthen any conclusions reached in the study of these relationships.
A possible third limitation is related to the generalizability of the sample. The
individuals who participated in this study come from a very specific industry. Although it is
102
not expected that the relationships in this study would be different in other occupations, it is a
concern worth mentioning. Given that these jobs are associated with caregiving, the
participants may represent a particularly altruistic or particularly conscientious portion of the
population in general. When the means for this sample are compared to the norms provided
by Costa and McCrae (1992), however, no practical differences exist. Nonetheless, it would
be well-founded to continue investigation of the findings reported here in other employee
samples.
Theoretical Implications
The most significant theoretical implications derived firom this study pertain to the
effectiveness of personality and affect in the prediction of non-workrole behaviors. In
particular, it is the relevance of personality to existing theories of organizational withdrawal
that is of greatest importance. As was stated in the introduction, theories of organizational
withdrawal have generally failed to include consideration of individual differences. To the
extent that individual differences have been considered, it has usually been limited to a very
specific manifestation such as preferences, needs, or experiences. The relationships
demonstrated in this study link withdrawal to an accepted general taxonomy of personality.
This link serves as an early step in establishing the nomological network around the areas of
personality and organizational behavior called for by Weiss and Adler (1984).
It has been established that personality was virtually abandoned as an explanatory
variable in I/O psychology for many years. The resxilts reported here contribute to the
continued interest in examining its usefulness in the study of organizational outcomes.
Personality has slowly increased its acceptance among researchers in the context of employee
selection and job performance. This study demonstrates that personality also has relevance to
the area of non-workrole behaviors such as organizational withdrawal and OCB.
The conceptualization of organizational withdrawal as a broad behavioral construct
played a key role in successfully demonstrating the relationship between personality and
103
behavior at work. By combining several forms of related criterion behavior, specific error
variance is minimized and variance that is common among the individual behaviors is
emphasized. It is very clear from the results of this study that the use of aggregation is
appropriate and beneficial to establishing relationships that have previously been overlooked.
A similar finding resulted from the use of criterion-related conceptualizations of
personality. Unlike attitudes and behavior where the conceptualization of behavior needed to
be broadened, the issue related to personality was the need to be more specific in selecting
what aspects of general personality are expected to be related to the criterion behaviors of
interest. The personality factors proved to be too broad and occluded their relationship with
the non-workrole behaviors. By looking at a factor's composition and selecting the portion of
the factor that is expected to be criterion-related, significant relationships were demonstrated
between personality and non-workrole behaviors.
Further evidence of the criterion-relatedness of the personality facets over the
personality factors is seen by comparing the amount of variance accounted for by the model
using personality factors and the variance accounted for by the model using personality
facets. The final model in series 1 (personality factors) accounted for 37% of the variance in
work withdrawal and 9% of the variance in OCB. Series 2 (personality facets) accounted for
43% of the variance in work withdrawal and 16% of the variance in OCB.
A theoretical issue that was not directly addressed in this study but is related to the
areas of job satisfaction and personality is the causal influence of personality on job
satisfaction. This is an area that provides opportunity for ftuther research. This line of
research has infrequently specifically addressed personality, and if personality was included,
it was not framed within the structure of the Big Five. In an attempt to address the issue
using this sample, a post hoc analysis was conducted to investigate a model in which the
personality variables were hypothesized to cause non-workrole behaviors through job
satisfaction in addition to the direct paths already specified.
104
Once again, two models were tested. Post hoc model 1 used the personality factors,
and post hoc model 2 used the personality facets. The direct paths remained the same for
both models. Post hoc model 1 did produce statistically significant path coefficients from
each of the factors to job satisfaction (see Figxure 18). The overall model fit statistics,
however, did not improve from those produced for Model 4 in series 1. The goodness of fit
indices for the first model are as follows: x~ ~ 547.75 with 269 degrees of freedom,
ratio = 2.04. GFI = .87, AGFI = .86, and the standardized RMSR = .08. The variance
accounted for in work withdrawal and OCB did not change from Model 4 in series 1. The
personality factors did, however, account for a substantial amount of variance in job
satisfaction (/?~ = .24) in the first post hoc model.
Post hoc model 2 consisted of the same analysis, only using personality facets in
place of the personality factors. This model produced generally weaker results (see Figure
19). Only impulsiveness and altruism produced significant path coefficients predicting job
satisfaction. The path from impulsiveness to job satisfaction was -.16 compared to -.42 for
neuroticism. The path from altruism to job satisfaction was .19 compared to.25 that was
produced by agreeableness. The overall model fit statistics did not improve over model 4
from the second series of models. The goodness of fit indices are as follows: x~ ~ 495.53
with 269 degrees of freedom, x'Idf ratio = 1.84, GFI = .88, AGFI = .87, and the standardized
RMSR = .07. Each of these values is equal the values reported for Model 4 in series 2.
However, the variance accounted for in job satisfaction by this model was .07, which is less
than in the first post hoc model.
Comparing the path coefficients and the variance accounted for by these two post hoc
models, once again, confirms the importance of choosing the proper conceptualization of the
Neuroticism i'acior
.19(1.79) - .S3
-.22(2.78) Job
Satisfaction -49
Conscientiousness
Factor -.19(3.09)
R' = .37 Work
Witiidrawal -.46(5.91)
.64 .25(4.7. .37
.08(.97)
Agreeableness Factor
• 37
-.43 - 4 5 .07(.81)
44 -.09(1.20) Negative
ATfect .02(.25
OrganizationalX R' - .09 Citizenship )
Behavior y - 3 5 .28(3.23)
i'ositive
A ffect
i-igure 18. Post hoc model showing personality factors causing job satisfaction. Completely standardized path coefllcienls are shown with t-values in parentheses.
Impulsiveness
I'acet
-.16(2.20)
-.30 /?- = .0
.01 (.09) Job
Satisfaction - .10
R' = .43 .19(1.97 J Work
Withdrawal -.54(6.41) 55
-.05(.74) Altruism
Facet - 3 7
-.22(1.97) -.27 - 2 3
Negative
A fleet .33(3.29)
Citizenship Behavior
- 3 4 -.31(3.64)
Positive
Affect
Figure 19. Post hoc model showing personality facets causing shown with 1-values in parentheses.
causing job satisfaction. Completely standardized path coefficients are
107
personality variables. In this case, the broader conceptualization of personality was a better
predictor of job satisfaction. Granted, these were post hoc analyses, so it would be
inappropriate to make definitive conclusions. The intent of the analyses, however, is to
provide direction for fiitiure research regarding the dispositional influences of job satisfaction.
The results presented here suggest that focusing on content-related conceptualizations of
personality is a key part of their continued investigation.
An additional post hoc analysis was conducted using the best fit models in both series
(Model 4) to fijrther investigate the relation between the two endogenous variables in the
original models. Given that the two concepts are relatively new, it is valuable to explore how
their relation affects the proposed models beyond what was originally hypothesized.
In this post hoc analysis, the residuals of the two endogenous variables were allowed
to correlate. In essence, this analysis provides information regarding the extent to which
OCB and work withdrawal are correlated after accounting for the intercorrelation of the
exogenous variables. In the original models, the two endogenous variables were not allowed
to correlate because they were hypothesized to be separate constructs. This hypothesis was
supported by the low zero-order correlation of the OCB and work withdrawal scales (r = -.06.
P > -05).
The post hoc analysis using the personalit>' factors produced interesting results (see
Figure 20). The overall fit of the model did improve over the hypothesized Model 4 in series
I. This post hoc model produced a decrease in chi-square that was statistically significant.
Ax" (1. ^V= 313) = 27.42,/? < .05. It also produced a x"/ df ratio of 1.94. The GFI increased
one point to .88. but the AGFI value did not change from .86. The RMSR decreased one
point to a value of .07. The estimate of variance accounted for only improved slightly to FT =
.39 for work withdrawal {Br was .37) while the variance accounted for in OCB remained the
same.
Neurotic ism
Factor
- 5 3 .21(2.03) = .39 - 5 0
Conscienliou.sness
^ I'actor > Work
Withdrawal .64 39 -.16(1.93
-.37 '-.42
-.21(3.34)
.10 - 4 2 -.44 .10(1.20)
.07(.81) Negative
Affect .37
Organizational
Citizenship
- 3 2 -.10(1.24)
Job
Satisfaction R- = .09 - 3 4 .24(2.93
Positive
Affect
Figure 20. Post hoc model using peisonality factors and allowing the work withdrawal and OC'Ii residuals to correlate. Cotiipletely standardized path coerftcients are shown with t-values in parentheses.
109
Most of the path coefficients in this new model increased slightly as a result of
correlating the endogenous variables. For example, the coefficient for neuroticism on work
withdrawal increased from .19 to .21. and the coefficient for conscientiousness on OCB
increased from .07 to .10. There was one coefficient however, that did decrease as a result of
allowing the variables to correlate. The path coefficient representing the relation between
positive affect and OCB decreased from .28 to .24.
When the same analysis was conducted using the personality facets (see Figure 21).
the overall model fit statistics again improved from the hypothesized Model 4. The chi-
square reduction was significant; Ax" (I, iV = 313) = 22.16, p < .05. The x'/ df ratio
decreased from 1.84 to 1.77. The GFI changed from .88 to .89 while the AGFI also increased
from .87 to .88. The RMSR decreased one point to .06.
Unlike the personality factors model, the individual relationships between the
variables were generally weaker than the relationships reported in the hypothesized model.
For example the coefficient for the path from job satisfaction to OCB changed from -. 13 to -
.11. and the coefficient for the path from dutifulness to work withdrawal changed from -.54
to -.53. In addition, there were relationships that changed considerably. The coefficient for
the path from dutifulness to OCB changed from being statistically significant at -.22. to being
non-significant at -.04. A similar decrease was found for the path between altruism and OCB
and the path between positive affect and OCB. It decreased from .31 to .24 but remained
statistically significant.
Another important difference from the personality factor model was the change in
variance accounted for. The personality facet model with the endogenous variables
correlated produced a decrease in variance accounted for in both work withdrawal and OCB.
The R~ for OCB decreased from .16 to .13 and the R' for work withdrawal decreased from .43
to .42.
Impulsiveness
Facet
-.30
Dutifulness
Facel
-.10 Work
Withdrawal
-.09(1.29) .25
Altruism Facel -.19(3.19) - 3 8
• 18
-.25 - 2 3
-.04(.38) Negative
Affect .26(2.69)
Organizational Citizenship
Behavior
-.32 -.11(1.56) 35
Job
Satisfaction - 3 4
.24(2.91)
At)
Positive
Affect
I'igiire 21. Post hoc model using personality facets and allowing the work withdrawal and OC'B residuals to correlate. Completely standardized path coeflkients are shown with t-values in parentheses.
I l l
The results of these post hoc analyses suggest that OCB and work withdrawal are
related and should be allowed to correlate in the models. This is inconsistent with the zero-
order correlations reported earlier. The multi-variate analysis revealed a relationship that was
obscured in the bi-variate analysis. The correlations estimated in the analyses were .29 in the
personality factors model and .27 in the personality facets model. Contrary to rational
expectation, OCB and work withdrawal were shown to be positively correlated. This result
suggests that there must be a common explanatory variable other than those included in the
models. For example, work load could be a possible determinant of both OCB and work
withdrawal and could be causing their positive correlation. Those employees who do not
have as much work to do have greater opportunities to engage in both positive and negative
non-workrole behaviors. This is obviously an interesting opportunity for further research.
Practical Implications
In addition to the theoretical implications already discussed, the results of this study
also have implications for the practice of I/O psychology. The renewed interest in
personality has mainly focused on its usefulness in employee selection. The results reported
here support the usefulness of personality post-hire as well. Work withdrawal, as a means of
removing one's self from work, presents an important performance management challenge.
Previously, work withdrawal would likely be addressed through designing an intervention
around the drivers of job satisfaction. Although this certainly is a valid approach, the
findings of this study would suggest that differences in personality may affect the success of
such an intervention. Given the personality traits of the individual are unlikely to be
effectively changed, knowledge of an individual's predisposition may influence the design of
the job satisfaction intervention to account for personality differences.
For example, suppose an organization has a work group where several of the workers
are coming in late and leaving early. When they are at work, many of them wander around
the work area and visit with their co-workers. In addition, recent changes in the company's
112
organization has resulted in significantly fewer supervisors in this particular area of
production. This fact prevents the organization from providing more supervision or
discipline. One option for an intervention might include trying to improve job satisfaction by
redesigning the job. Another option, based on the results of this study, would suggest
identifying those employees who have personality profiles that make them more likely to
engage in withdrawal behaviors (i.e.. those employees who are low in conscientiousness and
high in impulsiveness) and assign them to jobs that are less autonomous and have greater
supervision. Granted, there would be other factors to consider, such as employee relations
and the availability of other jobs, but given the right environment, such a strategy could be
beneficial.
Similar principles would apply to organizational managers wanting to increase the
occurrence of OCB. An example relevant to current management trends would be
organizations that are changing to a work structure that is based on teams from a structure
that is based on individuals. As responsibilities shift from individuals to teams, helping
behaviors that make up OCB will become increasingly important. Based on the results of
this study, consideration of individuals' altruism levels would be helpfixl in assembling the
teams and predicting the teams' success. Employees who are low on altruism could possibly
be put into jobs that are not team-based. Another option would be to include those
employees in the teams but alter the management of the team to account for the effects of low
altruism. Altering the management of such a team might include explicitly listing and
evaluating what are normally considered non-workrole behaviors. For example, including
the frequency of volunteering for projects or offering to help co-workers as part of
employees' performance appraisals.
The demonstrated influence of specific personality facets on work withdrawal and
OCB also presents an opportunity for continued research. This study examined a limited
number of possible personality facets that could be linked to non-workrole behaviors. A
113
possible next step would be to investigate other facets such as competence or aciiievement
striving from the conscientiousness factor that might be useful in the prediction of work
withdrawal and OCB.
The demonstrated relationship between positive affect and OCB has some
implications for performance management as well. Essentially, the results of this smdy say
that employees who have greater positive affect are more likely to engage in positive non-
workrole behaviors than employees who do not. This suggests that small gestures that
impact employees' mood may have an effect on employees' frequency of engaging in OCB.
A possible next step in this area would be to investigate what types of management activities
can improve or maintain employees' moods. For example, does casual day result in better
moods, and therefore, more OCB? Will bringing donuts in the morning result in more
frequent OCB?
The behaviors that compose OCB are generally considered trivial when taken
individually. However, when aggregated across time and across many individuals in an
organization, they can produce an important impact on organizational performance. To the
extent that employees' affective condition influences OCB performance, managers can
influence organizational outcomes by focusing on the mood and well-being of their
employees. The results reported here suggest that employees who are in a good mood are
more likely to volunteer, stay late, and/or make an extra effort. This provides
organizationally important reasons for supervisors and managers to be concerned with the
emotional state of their employees.
Conclusions
This smdy provided answers to questions regarding the usefulness of personality
measurement in the prediction of organizationally relevant employee behaviors. It provided
additional evidence to support the continued use of personality as an explanatory variable in
I/O psychology. In particular, it suggested that, like job attimdes. congruence between the
114
level or scope of measurement is an important consideration when establishing the usefulness
of personality as a predictor of employee behavior. Although personality has more or less
been accepted as consisting of five general factors, their general level of measurement is not
necessarily appropriate for all instances. In this study, evidence was provided that suggests
that decreasing the generality of the personality measurement and increasing the generality of
behavioral measurement demonstrates significant relationships between the two concepts.
Finally, the critique of personality by Guion and Gottier (1965) had profound
negative effects on personality research in applied settings. Things, however, have begun to
change as interest in the usefulness of personality continues to grow. Personality is now
conceptualized in a way that is very different and is producing results that are unlike the
results of 30 years ago. This study has contributed to the continued investigation of the new
understanding of personality, and how it can influence organizational effectiveness.
115
APPENDIX A The following questionnaire is being used to learn what it is like to work at
ORGANIZATION NAME, and how you feel about your job. The questionnaire will ask you about your attitudes, feelings, and behaviors regarding working at ORGANIZATION NAME. Instructions will be provided with each section. Questions are printed on both the FRONT AND BACK of each page. Some of the questions may seem similar to you, but this is necessary in order to provide a complete picture of your situation.
It is very Important that you answer each question as accurately as you can. There are no right or wrong answers. ALL ANSWERS WILL REMAIN CONFIDENTIAL. No one at ORGANIZATION NAME will see any one individual's answers. Please list the two job activities you do most often:
1.
2. Average number of hours worked per week
How long have you worked for ORGANIZATION NAME? Years Months
Circle the number corresponding to the appropriate answer.
Work Status at ORGANIZATION NAME; (1) Fulltime (2) Part time (3) Substitute
Sex: (1) Male(2) Female
Marital Status; (1) Married (2) Single (3) Divorced/Separated (4) Widowed
What is your current age?
(1) 18 to 23 (3) 30 to 35 (5) 42 to 47 (7) 54 to 58 (9) 65 and over
(2) 24 to 29 (4) 36 to 41 (6) 48 to 53 (8) 59 to 64 What is the highest level of education you have completed?
(1) Less than a high school education
(2) High school diploma or GEO (Graduate Equivalency Diploma)
(3) High school diploma plus some technical training or apprenticeship
(4) Some college
(5) Graduated from college (AA, BA, BS)
(6) Some graduate school
(7) Graduate or professional degree
(8) Other (please explain) What is your current salary?
(1) Below $15,000 (5) $45,001 - 55.000 (2) $15,001 -25,000 (6) $55,001 -65.000 (3) $25,001 - 35.000 (7) More than $65,000 (4) $35,001 - 45,000
Portions of the following are protected by copynght rules. Permission to use this sun/ey must be obtained from Kathy A. Hansen and/or PAR. inc. Portions of this survey are reproduced by speaal permission of the Publisher. Psychological Assessment Resources. Inc. 16204 North Flonda Avenue. Lutz. Florida 33549. from the NEO Five Factor Inventory, and NEO Pf-R by Paul Costa, and Robert McCrae. Copynght 1978. 1985. 1989. 1992 by PAR. Inc. Fuixher reproduction is prohibited without permission of PAR. Inc.
116
APPENDIX B
The following statements pertain to the WORK that you do. What is your WORK like MOST of the time? Please circle YES if the item describes your WORK, NO if the item does not describe your WORK, and circle ? only if you cannot dedde. Please circle one response for each question.
1. Fascinating 1. YES NO 7
2. Routine 2. YES NO ?
3. Satisfying 3. YES NO ?
4. Boring 4. YES NO ?
5. Important 5. YES NO ?
6. Creative 6. YES NO 7
7. Respected 7. YES NO ?
8. Pleasant 8. YES NO ?
9. Useful 9. YES NO 7
10. Tiresome 10. YES NO 7
11. Challenging 11. YES NO 7
12. Frustrating 12. YES NO 7
13. Simple 13. YES NO 7
14. Gives sense of accomplishment 14. YES NO 7
15. Dull 15. YES NO 7
16. A source of pleasure 16. YES NO 7
17. Awfiil 17. YES NO 7
18. Interesting 18. YES NO 7
In this section, you should think about your feelings about PROMOTION AND ADVANCEMENT in your job. All in ail, how do you feel about the promotion system? Circle YES if the item describes the PROMOTION system, NO if the item does not describe the PROMOTION system, and circle ? only if you cannot decide. Please circle one response for each item.
1. Good opportunity for advancement 1. YES NO 7
2. Opportunity somewhat limited 2. YES NO
3. Promotion on ability 3. YES NO 7
4. Dead-end job 4. YES NO 7
5. Good chance for promotion 5. YES NO 7
6. Infi-equent promotions 6. YES NO 7
7. Regular promotions 7. YES NO 7
8. Fairly good chance for promotion 8. YES NO 7
9. Easy to get ahead 9. YES NO 7
117
The next set of statements ask you to describe your IMMEDIATE SUPERVISOR. What is he/she like MOST of the time? Circle YES if the item describes your SUPERVISOR, NO if the item does not describe your SUPERVISOR, and circle ? only if you cannot decide. Please circle one response for each item.
1. Hard to please 1. YES NO ?
2. Impolite 2. YES NO ?
3. Praises good work 3. YES NO ?
4. Tactful 4. YES NO 7
5. Up-to-date 5. YES NO ?
6. Quick-tempered 6. YES NO ?
7. Tells me where 1 stand 7. YES NO ?
8. Annoying 8. YES NO ?
9. Stubborn 9. YES NO ?
10. Knows job well 10. YES NO ?
11. Bad 11. YES NO ?
12. Intelligent 12. YES NO ?
13. Lazy 13. YES NO ?
14. Around when needed 14. YES NO ?
15 Interferes with my work 15. YES NO ?
16. Gives confusing directions 16. YES NO ?
17. Knows how to supervise 17. YES NO ?
18. Cannot be trusted 18. YES NO ?
The following questions pertain to the PAY AND BENEFITS you receive from your job. Please circle YES if the item describes your PAY AND BENEFITS, NO if the item does not describe your PAY AND BENEFITS, and circle ? only if you cannot decide. Please circle one response for each question.
1. Income adequate for normal expenses 1. YES NO ?
2. Barely live on income 2. YES NO 7
3. Bad 3. YES NO 7
4. insecure 4. YES NO ?
5. Less than 1 deserve 5. YES NO 7
6. Underpaid 6. YES NO 7
7. Well paid 7. YES NO 7
8. Unfair 8. YES NO 7
9. Enough for what 1 need 9. YES NO 7
118
The following statements ask you to think about the majority of the employees you work with. What are they like MOST of the time? Circle YES if the item describes the PEOPLE YOU WORK WITH, NO if the item does not describe the PEOPLE YOU WORK WITH, and circle ? only if you cannot decide. Please circle one response per item.
1. Stimulating 1. YES NO ?
2. Boring 2. YES NO ?
3. Slow 3. YES NO ?
4. Ambitious 4. YES NO ?
5. Stupid 5. YES NO ?
6. Responsible 6. YES NO ?
7. Waste of time 7. YES NO ?
8. Intelligent 8. YES NO ?
9. Easy to make enemies 9. YES NO ?
10. Talk too much 10. YES NO ? 11. Smart 11. YES NO ?
12. Lazy 12. YES NO ? 13. Unpleasant 13. YES NO ? 14. Active 14. YES NO ?
15 Narrow interests 15. YES NO ?
16. Loyal 16. YES NO ?
17. Bother me 17. YES NO ?
18. Work well together 18. YES NO ?
119
APPENDIX C
Neuroticism
I am not a worrier. (R) I often feel inferior to others. When I'm under a great deal of stress, sometimes I feel like I'm going to pieces. In dealing with other people, I always dread making a social blunder. I often feel tense and jittery. Sometimes I feel completely worthless. I rarely feel fearfiil or anxious. (R) I often get angry at the way people treat me. Too often, when things go wrong, I get discouraged and feel like giving up. I am seldom sad or depressed. (R) I often feel helpless and want someone else to solve my problems. At times I have been so ashamed I just wanted to hide.
Agreeableness
I try to be courteous to everyone I meet. I often get into arguments with my family and co-workers. (R) Some people think I'm selfish and egotistical. (R) I would rather cooperate with others than compete with them. I tend to be cynical and skeptical of others' intentions. (R) I believe that most people will take advantage of you if you let them. (R) Most people I know like me. Some people think of me as cold and calculating. (R) I'm hard-headed and tough-minded in my attimdes. (R) I generally try to be thoughtful and considerate. If I don't like people. I let them know it. (R) If necessary. I am willing to manipulate people to get what I want. (R)
Conscientiousness
I keep my belongings neat and clean. I'm pretty good about pacing myself so as to get things done on time. I am not a very methodical person. (R) I tiy to perform all the tasks assigned to me conscientiously. I have a clear set of goals and work toward them in an orderly fashion. I waste a lot of time before settling down to work. (R) I work hard to accomplish my goals. When I make a commitment, I can always be counted on to follow through. Sometimes I'm not as dependable or reliable as I should be. (R) 1 am a productive person who always gets the job done. I never seem to be able to get organized. (R) I strive for excellence in everything I do.
Note: R indicates an item diat is reverse scored.
120
APPENDIX D
Neuroticism - Facet N5: Impulsiveness
I rarely overindulge in anything. (R) I have trouble resisting my cravings. I have little difEcxilty resisting temptation. (R) When I am having my favorite foods, I tend to eat too much. I seldom give in to my impulses. (R) I sometimes eat myself sick. Sometimes I do things on impulse that I later regret. I am always able to keep my feelings under control. (R)
Agreeableness - Facet A3: Altruism
Some people think I'm selfish and egotistical. (R) I try to be courteous to everyone I meet. Some people think of me as cold and calculating. (R) I generally try to be thoughtful and considerate. I'm not known for my generosity. (R) Most people I know like me. I think of myself as a charitable person. I go out of my way to help others if I can.
Conscientiousness - Facet C3: Dutifiilness
I tr>- to perform all the tasks assigned to me conscientiously. Sometimes I'm not as dependable or reliable as I should be. (R) I pay my debts promptly and in fiill. Sometimes I cheat when I play solitaire. (R) When I make a commitment, I can always be counted on to follow through. I adhere strictly to my ethical principles. 1 try to do jobs carefiilly, so they won't have to be done again. I'd really have to be sick before I'd miss a day of work.
Note: R indicates an item that is reverse scored.
121
APPENDIX E
The following items ask you how characteristic each of the following behaviors are of your work activities. Indicate how much you agree with each statement by writing the number of the response in the blank next to each item. Please use the response scale below. Again, all answers are confidential.
1 2 3 4 5 Strongly Disagree Neutral Agree Strongly Agree Disagree
c1. I help other employees with their work when they have been absent.
c2. I take initiative to orient new employees even though if s not part of my job description.
w3. I sometimes take undeserved work breaks.
c4. I take fewer days off than other employees.
c5. I give advance notice if I'm unable to come to work.
w6. I coast toward the end of the day.
c7. I willingly attend wori< functions that are not required, but are good for the company.
•Vbre: c indicates citizenship item and w indicates a work withdrawal item.
122
APPENDIX F
The following items ask you to estimate how frequently you have engaged in certain job behaviors or how often various activities have occurred in the past 12 months. Please use the response scale below and write the number of the response in the blank next to each item. Again, all answers are confidential.
1 Never
2 Maybe once a
year
3 Two or
three times a year
4 Nearly
every other month
About once a month
6 More than
once a month
7 Once a week
8 More than
once a week
w
w
w
w
w
w
w
w
w
w
w
_ Drinking alcohol or using drugs after work primarily because of things that occurred at work.
Doing things that are not required on my job that make ORGANIZATION NAME a better place to work.
Finding easier work to avoid doing unpleasant tasks (such as changing linens, bed pans, etc.).
. Talking up ORGANIZATION NAME to my friends as a great organization to work for.
Not completing required paperwork/charting on time.
. Arriving at work before my scheduled time in the morning.
Taking frequent or long coffee or lunch breaks.
Staying late to help a co-worker even when i would not have to.
Helping others when their wori< load increases.
Drinking alcohol or using drugs before coming to work.
. Volunteering to do things not formally required of my job.
Neglecting those tasks that will not affect my performance evaluation.
Taking extra care with equipment to keep it in good shape.
Letting others do my work for me during my shift.
. Withholding important information fi-om co-wori<ers or supervisors.
Making suggestions to my supervisor about better ways to do things at ORGANIZATION NAME.
Making up excuses to get out of going to wori«.
Making an extra effort to keep things neat, clean, and orderiy.
Complaining to other employees about ORGANIZATION NAME.
Starting work eariy after returning from lunch and/or breaks.
w Visiting with co-workers about trivia while at work.
c Purposefully leaving work for the next shift to do.
w Taking responsibility for initiating needed changes in my wotlc.
w Being absent when I am not actually sick.
w Violating a company safety rule.
w Daydreaming while I should be working.
c Making suggestions to improve quality.
w Making excuses to leave the work area.
w Failing to attend scheduled meetings.
w Gossiping with others at work.
c Assisting my supervisor with his/her duties.
w Using the woric phone for personal calls.
w Leaving work eariy without permission.
c Working on more difficult tasks to make things better for the next shift.
Note: c indicates citizenship item and vv indicates a worlc withdrawal item.
123
APPENDIX G
The PANAS
This scale consists of a number of words that describe different feelings and emotions. Read each item and then mark the appropriate answer in the space next to that word. Indicate to what extent you have felt this way during the PAST V\teEK across all areas of your life. Use the following scale to record your answers.
1 2 3 4 5 very slightly or a little moderately quite a bit extremely
not at all
. interested irritable
. distressed alert
. excited ashamed
. upset inspired
. strong nervous
. guilty detennined scared attentive hostile jittery enthusiastic active proud afraid
124
APPENDIX H DATE
Dear ORGANIZATION employee:
As organizational researchers at Iowa State University, we are interested in studying employee behavior in different types of jobs. As a result, we are requesting your participation in a research project that is being conducted at ORGAMZATION. This study will examine how employees react to certain aspects of their jobs.
Your participation in the project is voluntary and consists of answering questions about yourself, how you feel about your job, and the different types of behaviors you might engage in at work. It will take approximately 35-40 minutes to complete the questiormaire. The project has been approved by the administrators at ORGANIZATION and they are encouraging your participation.
The information you provide will be kept absolutely confidential. Actual responses will not be seen by anyone but the researchers. The questionnaires will be returned directly to Iowa State University via a pre-paid, pre-addressed envelope included with the questiormaire. No names are needed for Ae questionnaire. ORGANIZATION will receive a report of the smdy's findings, but no individual will be able to be identified because all information in the report will be combined into groups. This means that no one at ORGANIZATION will see your answers or be able to identify an employee from the information gathered in this study. After the questionnaires have been received, they will be kept in a locked file at Iowa State University.
The purpose of this project is to provide information on the relationship between employee attitudes, personality, and behavior at work. Conclusions drawn from this project are expected to be useful to organizations in general as well as care providers such as ORGANIZATION. The information gathered here may be used in scientific presentations and publications, but only after it has been combined into groups and all personal identifying information has been removed.
Everyone at ORGANTZATION is receiving a questionnaire. For this project to be successful, we must receive responses from as many ORGANIZATION employees as possible. In appreciation of your completing and returning the questiormaire by DATE, your name will be entered into a drawing for $100. Simply fill out the enclosed form and return it in the prepaid envelope with your questionnaire. The winner will be notified by phone during the month of July. If you are not comfortable with including the drawing form in the envelope with your questionnaire, you may mail it separately to Doug Molitor at the address above. If you have any questions about this study, please feel free to call Doug Molitor at 703-351-9407. Thank you for your time and participation.
Sincerely,
Douglas Molitor Iowa State University
Kathy Hanisch. Ph. D. Iowa State University
APPENDIX I Correlations Among Manifest Indicators
NIA N2A N3A AGGl AGG2 AGG3 CONl CON 2 CON 3 IMPl IMP2
NIA 1.0000 N2A .5507 1.0000 N3A .5479 . 5387 1.0000 AGGl - .2409 -.2711 -.1108 1.0000 AGO 2 -.3151 -.3062 - . 1318 .3547 1 , 0000 AGG3 -.2817 -.2976 -.1708 .4120 ,5370 1 . 0000 CONl -.2754 -.2982 -.2028 .2314 ,0958 .2645 1,0000
CON 2 -.3076 -.3106 -.2431 .2196 ,0407 . 1890 ,5053 1.0000 CON 3 -.3885 -.3213 -.2279 .2799 .0964 .2682 ,5591 .5889 1,0000
IMPl .3139 .3860 .2585 -.1747 - , 1267 - . 1879 - , 1210 -.0887 -,1507 1,0000 IMP2 .3985 .6288 . 3474 -.2181 - . 1646 - .2150 - ,2131 - .2439 -,2767 . 5122 1 ,0000 IMP3 .2879 .3788 .2321 -.0147 -.0898 -.1663 -,0866 - .0819 -,0874 .3518 ,3764 ALTl -.1758 -.1396 - . 1044 .3739 .3316 .6648 ,2737 .2131 .2603 - .0906 -,0740 ALT2 -.2048 -.1480 -.1346 .5654 .2231 .3116 . 1500 .2661 .2065 - .0421 -.0480 ALT 3 -.1769 -.1310 -.1194 .3875 .4302 .4659 , 1875 .2376 .1953 -.0364 -,0635 DUTl -.2532 -.2258 -.1428 .3152 .1518 .2285 ,3518 . 5860 .4779 -.1326 -.2096 DUT2 -.2859 -.2745 -.2083 . 1544 .0433 . 1475 ,2549 . 5604 .4178 -.0641 - . 1738 DUT3 -.2048 -.2688 -.1677 .2155 . 1710 .1722 ,2107 . 3592 .2156 -.0966 - . 1266 POSAFFl -.2894 -.3854 -.2136 .2380 .2241 . 1798 , 1944 .3518 .2639 -.1538 - . 1956 P0SAFF2 -.2023 -.2555 -.1663 .1925 . 0537 .0968 .2310 . 3626 .2362 -.0703 - . 1267 P0SAFF3 -.3137 -.3494 -.2361 , 1505 .0993 , 1565 .2746 . 3000 ,2518 -.1028 - .2050 NEGAFFl .4121 . 3185 ,4360 - . 1631 - .2716 -.2299 -,1602 -.1533 -.2457 . 1045 . 1568 NEGAFF2 .4445 . 3634 .3321 - .3058 - .2986 -.2713 - .2348 -.1419 -.2962 .1197 . 1569 NEGAFF3 .3950 . 3699 .4037 - .2116 -.2154 -.2031 - .2637 -.2065 - .3218 .1333 . 2063 JDIl -.3221 - .2881 -.2264 . 1726 .3073 . 1473 - .0442 .0951 . 0680 -.0946 -.1111 JDI2 -.3369 -.3146 -.2020 .2394 .3206 . 1863 .0152 .1165 .0930 - . 1496 -.1401 JDI3 -.3190 -.3100 -.2135 .2156 .3427 .2153 .0103 .1117 .0962 - , 1340 -.1565 WWl . 3288 . 3646 .1816 - . 1527 - .2522 -,1884 -.2602 -.2308 -,2684 ,1284 .2138 WW2 .3020 . 3780 .2083 -.2404 - .2344 -,3026 -.3397 -.3531 -.3382 , 1543 .2894 WW 3 . 2333 . 2587 .1187 -,2178 -.1858 -,3026 - .4124 -,3382 - . 3631 , 1276 .2040 OCBl -.0505 -.0637 -.0734 , 1603 -.0580 , 1048 . 1209 ,2039 .1045 -,0477 -.1123 0CB2 -.1567 -.2103 - . 1779 , 2035 .0417 , 1497 .2155 ,2723 . 1277 -,0964 - . 1706 0CB3 -.0073 .0211 -.0402 , 1002 - . 1048 , 0483 .1162 , 1581 . 0712 , 0039 - . 0313
Appendix 1 Continued.
IMP3 ALT! ALT2 ALT3 DUTl DUT2 DUT3 POSAFFl P0SAFF2 POSAFF3 NEGAFFl
NIA N2A N3A AGGl AGG2 AGG3 CONl CON 2 CON 3 IMPl IMP2 IMP3 1.0000 ALTl -.0663 1.0000 ALT2 . 0667 .4277 1.0000 ALT 3 .08B7 .4452 .4390 1.0000 DUTl . 0533 .1991 .3400 .2824 1.0000 DUT2 - . 0569 .1162 .1655 .1143 .3325 1 . 0000 DUT3 . 2487 .0994 .2364 . 1745 .2941 ,2907 1.0000 POSAFFl . 0252 . 1724 .2905 . 1985 .2915 . 1682 .2998 1.0000 P0SAFF2 .0199 .1348 .2618 . 1624 .1824 ,1454 ,2352 .6957 1.0000
P0SAFF3 -.0343 .0726 .2194 .1243 . 1707 .1543 .1983 .6955 .6734 1.0000 NEGAFFl .0766 -.1441 - .0520 -.1844 -.1808 - . 1525 -,1404 - .2125 - .0927 -,2127 1 .0000 NEGAFF2 .1379 -.1531 - .1655 -.1937 -.1680 -.1250 -,0984 - .3095 -.2526 - , 3440 . 6472 NEGAFF3 . 1057 -.1647 - .1169 -.1572 -.2400 -.1656 -,1939 - .2462 -.2127 -,2506 .6921 JDIl -.0719 . 0582 .0965 . 1482 .0800 .0524 , 1044 .3814 .3307 ,2975 -.2304 JDI2 - . 0317 . 1088 .1781 . 1768 . 1081 . 0487 ,1085 .4062 .3834 ,3275 -.2156 JDI3 - .0823 . 1220 .1231 . 1386 ,0885 .0657 .0862 .3873 . 3280 .3367 - .2398 WWl . 1247 -.1507 -.1181 - . 1828 - . 1410 -.1850 - . 1898 -.2705 -.2283 - .2575 .2234 WW2 .1330 -.2208 -.1967 -.2037 -.2610 - . 3089 -.3308 -.2780 -.2132 - . 2258 . 1564 WW 3 . 1391 -.2776 - . 1841 -.2094 -.2006 -.3059 -.3061 -.1686 -.1390 - . 1463 . 1289 OCBl .0704 .1636 .2171 . 1348 . 1689 .0170 .0401 .2004 .2020 .1521 . 0569 OCB2 .0144 . 1629 .2303 . 1447 .2345 .1332 .1295 .2891 .3084 .2839 . 0029 0CB3 .0577 . 0939 . 1645 . 0626 , 1084 .0362 .0966 .0788 . 1520 . 1009 , 0948
Appendix I Continued.
NEGAFF2 NEGAFF3 JDIl JDi:: JDI3 WWl WW2 WW3 OCBl 0CB2 0CB3
NIA
N2A
N3A
AGGl
AGG2
AGG3
CONl
CON 2
CON 3
IMPl
IMP2
IMP3
ALTl
ALT2
ALT3
DUTl
DUT2
DUT3
POSAFFl P0SAFF2 P0SAFF3 NEGAFFl NEGAFF2 1 .0000 NEGAFF3 .6182 JDIl - .3178 JDI2 -.3464 JDI3 - . 3540 WWl .3374
WW2 .2333 WW 3 . 1747 OCBl .0612 0CB2 -.0605 0CB3 .1081
. 0000
.1901 1 . 0000
. 1859 .8817
. 1865 .8602
. 1840 - .3685
.1337 -.2373
. 1713 -.1055
.0383 .0162
. 0802 . 1257
. 0685 -.0986
I.0000 .8795 1 .0000
- .3243 - . 3293 - .2404 -.2374 -.1320 -.1025 . 0727 . 0426 . 1 774 .1719
-.0243 -.0790
0000 5970 1.0000 4946 . 7066 2259 . 1098 1625 - . 1046 3390 . 0609
1 . 0 0 0 0 .0796 1.0000
-.0160 .6146 .1188 .6926
1 . 0 0 0 0 .5659 1.0000
APPENDIX J Covariance Matrix Used In Personality Factor Modeling Procedures
WWl WW2 WW3 OCBl OCB2 0CB3
WWl 32 . 85 WW2 20 . , 13 34 . 61 WW3 16 . , 91 24 . 80 35 . 59
OCBl 7 , , 79 3 . , 89 2 . ,86 36 . , 20 0CB2 6 , ,64 -4 , ,39 -0 , ,68 26, ,36 50. 83 0CB3 9 , , 19 1 . , 70 3 . , 35 19 . ,71 18 . 75 22 . 38 NIA 5 . , 19 4 , , 89 3 , ,83 -0 , , 84 "3 . 07 -0 . 09 N2A 5 . , 64 6 . 00 4 . , 17 -1. , 04 -4 . 05 0 . 27 N3A 2 , , 73 3 , ,22 1 . , 86 -1. , 16 -3 . 33 -0 . , 50
AGGl -1 . .44 -2 , , 33 -2 . , 14 1, ,59 2 , ,39 0 , ,78 AGG2 -3 . . 59 -3 , ,43 -2 , , 75 -0 , , 87 0 , , 74 -1 , , 23 AGG3 -2 , .38 -3 . , 92 -3 , ,98 1 , .39 2 , ,35 0 , , 50 CONl -3 , . 09 -4 , . 14 -5 . , 10 1 . .51 3 , , 19 1 , , 14 C0N2 -2 . 65 -4 , . 16 -4 , . 04 2 , .46 3 . , 89 1 . . 50 C0N3 -3 .28 -4 , .25 -4 . . 63 1 . . 34 1 . , 94 0 . , 72
POSAFFl -4 . 37 -4 , . 61 -2 , . 83 3 , .40 5 , . 80 1 . . 05 P0SAFF2 -2 . 87 -2 .75 -1 .82 2 .66 4 , . 82 1 . 58 P0SAFF3 -3 . 16 -2 . 85 -1 , .87 1 . 96 4 , . 34 1 , . 02 NEGAFFl 3 . 32 2 . 38 1 . 99 0 . 89 0 , . 05 1 . 16 NEGAFF2 3 .48 2 .47 1 . 88 0 .66 -0 , .78 0 . 92 NEGAFF3 2 . 10 1 .57 2 .03 0 .46 -1 . 14 0 .65
JDIl -25 . 24 -16 . 69 -7 . 52 1 . 16 10 .71 -5 .58 JDI2 -22 . 30 -16 . 97 -9 .45 5 . 25 15 . 17 - 1 .38 JDI3 -24 .27 -17 . 96 -7 . 86 3 .30 15 . 76 -4 .80
Covariance Matrix Continued
NIA N2A N3A AGGl AGG2 AGG3
NIA 7. 57 N2A 4 . 09 7 . 29 N3A 3 . 96 3 . 82 6 . 89
AGGl -1 . 09 -1 . 21 -0 . 48 2 . 72 AGG2 -2 . 15 -2 . 05 -0 . 86 1 . 45 6 . 17 AGG3 -1 . 71 -1 . 77 -0 . 99 1 . 50 2 . 94 4 . 85 CONl -1 . 57 -1 . 67 -1 . 10 0 . 79 0 . 49 1 . 30 C0N2 -1 . 70 -1 . 68 -1 . 28 0 . 73 0 . 20 0 . 83 C0N3 -2 , ,28 -1 , ,85 -1 , ,28 0, ,99 0. 51 1 . 26
POSAFFl -2 , , 24 -2 . , 93 -1 . , 58 1 . , 11 1. 57 1 . 12 P0SAFF2 -1 , , 22 -1 , , 51 -0 . , 96 0 . ,70 0 . 29 0 . 47 P0SAFF3 -1. , 85 -2 , , 02 -1 , ,33 0 , ,53 0 , 53 0 . 74 NEGAFFl 2 , . 94 2 , ,23 2 , . 96 -0 , ,70 -1. 75 -1 . , 31 NEGAFF2 2 , .20 1. . 77 1 , , 57 -0 , , 91 -1 . 34 -1 . 08 NEGAFF3 2 , , 16 1. . 99 2 . , 11 -0 . ,69 -1. 07 -0 . , 89
JDIl -10 . . 60 -9 , . 30 -7 , , 11 3 , .40 9 . , 13 3 , , 88 JDI2 -11 . . 12 -10 , . 19 -6 , ,36 4 . , 74 9 . 56 4 , , 92 JDI3 -11 , .29 -10 . 77 -7 , .21 4 , . 57 10 , , 95 6 , , 10
CONl C0N2 C0N3 POSAFFl P0SAFF2 POSAFl
CONl 4 .30 C0N2 2 . 10 4 . 01 C0N3 2 .48 2 . 52 4 . 56
POSAFFl 1 . 14 1 . 98 1 . 59 7 . 93 P0SAFF2 1 . 05 1 . 59 1 . 11 4 .29 4 , . 81 POSAFF3 1 . 22 1 . 29 1 . 15 4 .20 3 . 16 4 . 59 NEGAFFl -0 . 86 -0 . 80 -1 . 36 -1 .55 -0 . 53 -1 . 18 NEGAFF2 -0 . 88 -0 . 51 -1 . 14 -1 . 57 -1 . 00 -1 . 33 NEGAFF3 -1 . 09 -0 . 82 -1 . 3 7 -1 .38 -0 . 93 -1 . 07
JDIl -1 . 10 2 . 28 1 .74 12 . 84 8 . 66 7 .62 JDI2 0 . 38 2 . 80 2 .38 13 . 72 10 . 08 8 .42 JDI3 0 .28 2 . 88 2 . 64 14 .02 9 . 25 9 .28
Covariance Matrix Continued
NEGAFFl NnGAFF2 NF.GAFF3
NEGAFFl 6 .71 NEGAFF2 3 . 02 3 . , 26 NEGAFF3 3 . 57 2 . . 22 3 , . 96
JDIl -7. 14 -6 , , 85 -4 , . 52 JDI2 -6 . 70 -7 . .49 -4 , .44 JDI3 -7.99 -8 , . 21 -4 . 77
JDll JDI2 JD13
142 . 89 126.44 143.94 132.24 135.70 165.40
APPENDIX K Covariance Matrix Used In Personality I'acet Modeling Procedures
WWl WW2 WW3 OCBl OCB2 0CB3
WWl 32 . 85 WW2 20 . 13 34 . 61 WW3 16 . , 91 24 , , 80 35 . 59
OCBl 7 . , 79 3 . ,89 2 . 86 36 . , 20 0CB2 6 , , 64 -4 , ,39 -0 . 68 26 , .36 50 . ,83 0CB3 9 , , 19 1 , ,70 3 , ,35 19. .71 18 . , 75 22 . , 38 IMPl 1 , . 11 1 , .37 1 . , 15 -0 , ,43 -1 . 03 0 . 03 IMP2 2 , .29 3 , . 18 2 , , 27 -1 , . 26 -2 , . 27 -0 . ,28 IMP3 1 . , 14 1 . . 24 1 , , 32 0 , . 67 0 , , 16 0 . ,43 ALTl -1 . , 35 -2 . . 03 -2 , , 59 1, . 54 1 . . 82 0 . .70 ALT2 -0 , . 93 -1 , . 59 -1 . , 51 1 , . 80 2 , . 26 1 , , 07 ALT3 -1 . , 16 -1 , .33 -1 . ,39 0 , . 90 1 . , 14 0 . ,33 DUTl -1 . 24 -2 , .35 -1 . . 83 1, . 55 2 . . 56 0 , . 79 DUT2 -1 .75 -3 , . 00 -3 , . 01 0, . 17 1 , . 57 0 , . 28 DUT3 -1 .63 -2 . 92 -2 , . 74 0 . 36 1 , , 39 0 , . 69
POSAFFl -4 . 37 -4 . 61 -2 , . 83 3 , .40 5 , . 80 1 , . 05 P0SAFF2 -2 .87 -2 .75 -1, . 82 2 , .66 4 , . 82 1 , .58 P0SAFF3 -3 . 16 -2 . 85 -1 . 87 1 . 96 4 , . 34 1 , . 02 NEGAFFl 3 . 32 2 .38 1, . 99 0 , . 89 0 , . 05 1 , . 16 NEGAFF2 3 .48 2 .47 1 .88 0 .66 -0 . 78 0 . 92 NEGAFF3 2 . 10 1 . 57 2 . 03 0 .46 -1 , . 14 0 .65
JDIl -25 . 24 -16 . 69 -7 , . 52 1, . 16 10 , . 71 -5 , . 58 JDI2 -22 .30 -16 . 97 -9 .45 5 . 25 15 . 17 -1 . 38 JDI3 -24 . 27 -17 . 96 -7, . 86 3 .30 15 . 76 -4 . 80
Covariance Matrix Continued
IMPl 1MP2 1MP3 ALTl ALT2 ALT3
IMPl 2 . 26 IMP2 1. ,44 3 . 48 IMP3 0 . , 84 1. 12 2 . 53 ALTl -0 . ,21 -0 . ,22 -0 . 17 2 , ,45 ALT2 -0 , ,09 -0 . ,12 0 . , 15 0 , , 92 1. 89 ALT3 -0 , , 06 -0 , , 13 0 , , 16 0 , ,77 0 . 67 1 . ,23 DUTl -0 , . 31 -0 . ,60 0 , , 13 0 , ,48 0 . ,72 0 . ,48 DUT2 -0 . . 16 -0 , , 53 -0 , , 15 0 , , 30 0 . , 38 0 , , 21 DUT3 -0 , , 22 -0 . .35 0 , , 59 0 . ,23 0 , ,49 0 , ,29
POSAFFl -2 , . 24 -2 , . 93 -1 , , 58 1, , 11 1 . , 57 1. , 12 P0SAFF2 -1, .22 -1 , . 51 -0 . . 96 0 , , 70 0 , , 29 0 , .47 P0SAFF3 -1, .85 -2 . 02 -1 , . 33 0 , . 53 0 , , 53 0 . , 74 NEGAFFl 2 .94 2 .23 2 , , 96 -0 .70 -1 , .75 -1, . 31 NEGAFF2 2 . 20 1 . 77 1, . 57 -0 . 91 -1 , .34 -1, . 08 NEGAFF3 2 .16 1 .99 2 . 11 -0 .69 -1 , , 07 -0 . 89
JDIl -10 . 60 -9 . 30 -7 . 11 3 .40 9, . 13 3 . 88 JDI2 -11 . 12 -10 . 19 -6 .36 4 . 74 9 . 56 4 . 92 JDI3 -11 .29 -10 .77 -7 , . 21 4 .57 10 , , 95 6 . 10
DUTl DUT2 DUT3 POSAFFl P0SAFF2 P0SAFF3
DUTl 2 , , 34 DUT2 0 , , 84 2 , ,72 DUT3 0 , .68 0 , ,72 2 . 25
POSAFFl 1 , . 26 0 , , 78 1 , ,27 7 , , 93 P0SAFF2 0, . 61 0 , , 53 0 , ,77 4 , , 29 4 . , 81 P0SAFF3 0 , . 56 0 , . 55 0 . . 64 4 . , 20 3 , , 16 4 , , 59 NEGAFFl -0 , , 72 -0 . , 65 -0 , , 55 -1 , , 55 -0 . , 53 -1 , , 18 NEGAFF2 -0 , .46 -0 , , 3 7 -0 , . 27 -1 , . 57 -1 . , 00 -1 , , 33 NEGAFF3 -0 , . 73 -0 , , 54 -0 , , 58 - 1 , , 38 -0 . , 93 -1 . , 07
JDIl 1 .46 1 . 03 1 . 87 12 . 84 8 , . 66 7 . 62 JDI2 1 . 99 0 , . 96 1 . 96 13 .72 10 . 08 8 .42 JDI3 1 . 74 1 .39 1 . 67 14 . 02 9 . 25 9 .28
Covariance Matrix Continued
NEGAFFl NnGAFF2 NRGAFF3
NEGAFFl 6 , , 71 NEGAFF2 3 . . 02 3 , , 25 NEGAFF3 3 , , 57 2 . 22 3 , . 96
JDIl -7 , , 14 -6 , . 85 -4 , . 52 JDI2 -6 , , 70 -7 . ,49 -4 , ,44 JDI3 -7 . 99 -8 , . 21 -4 , . 77
JDIl JDI2 JDI3
142.89 126.44 143.94 132.24 135.70 165.40
134
REFERENCES CITED
Aamondt M. G. (1996). Applied Industrial/Organizational Psychology. (2nded.). New-
York: Brooks/Cole Publishing.
Abrahams, N. M., Neumann, L, & Githens. W. H. (1971). Faking vocational interests;
Simulated versus real life motivation. Personnel Psychology, 24. 5-12.
Adomo, T. W., Frenkel-Brunswik, E., Levinson, D. J.. & Sandford. N. (1950). The
Authoritarian Personality. New York: Harper & Row.
Ajzen, I. (1988). Attitudes. Personality, and Behavior. Chicago: Dorsey Press.
Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human
decision Processes. 50. 179-211.
Ajzen. I. & Fishbein, M. (1977). Attitude-behavior relations: A theoretical analysis and
review of empirical research. Psychological Bulletin. 84(5). 888-918.
Allport, G. W. & Odbert, H. S. (1933). Trait-names: A psycho-lexical study. Psychological
Monographs. 47. 171-220 (1, Whole No. 211).
Anderson. J. C. & Gerbing, D. W. (1988). Structural equation modeling in practice: A
review and recommended two-step approach. Psychological Bulletin. 103(3). 411-
423.
Arvey. R. D.. Bouchard. T. J., Jr.. Segal. N. L., & Abraham, L. M. (1989). Job satisfaction:
Envirorunental and genetic components. Journal of Applied Psychology. 74. 187-
192.
Atkinson. J. W. & Birch. D. (1970). The Dynamics of Action. New York: Wiley.
Barrick. M. R. & Mount, M. K. (1991). The big five personality dimensions and job
performance: A meta-analysis. Persormel Psychology. 44. 1-26.
Barrick, M. R. & Mount. M. K. (1993). Autonomy as a moderator of the relationships
between the big five personality dimensions and job performance. Journal of Applied
Psychology. 78. 111-118.
135
Bateman. T. S. & Organ, D. W. (1983). Job satisfaction and the good soldier: The
relationship between affect and employee citizenship. Academy of Management
Journal. 26(4). 587-595.
Baumgarten, F. (1933). Die Charaktereigenshaften. Beitrage zur Charakter und
Personlichkeitsforshimg: Monogr. I. Bern: A. Francke.
Bentler. P. M. & Chou, C. (1987). Practical issues in structural modeling. Sociological
Methods and Research. 16. 78-117.
Bemardin, H. J. (1977). The relationship of personality variables to organizational
withdrawal. Personnel Psychology. 30. 17-27.
Bemardin, H. J. & Bownas, D. A. (Eds.). (1985). Personality Assessment in Organizations.
Block. J. (1993). A contrarian view of the five-factor approach to personality description.
Unpublished manuscript. University of California, Department of Psychology.
Berkeley.
Borgatta. E. F. (1964). The structure of personality characteristics. Behavioral Science. 9,
8-17.
Borman. W. C., & Motowidlo. S. J. (1993). Expanding the criterion space to include
elements of contexmal performance. In N. Schmitt & W, C. Borman (Eds.),
Personnel Selection (pp. 71-98). San Francisco, CA; Jossey-Bass.
Brayfield. A. H.. & Crockett, W. H. (1955). Employee attitudes and employee performance.
Psychological Bulletin. 52. 396-424.
Brief. A. P.. Burke. M. J., George, J. M., Robinson, B. S., & Webster. J. (1988). Should
negative affectivity remain an uruneasured variable in the smdy of job stress? Journal
of Applied Psychology. 73(2). 193-198.
Brief. A. P. & Motowidlo, S. J. (1986). Prosocial organizational behaviors. Academy of
Management Review. 11(4). 710-725.
136
Brief. A. P. & Roberson. L. (1989). Job attitude organization: An exploratory study.
Journal of Applied Social Psychology. 19. 711-111.
Cattail. R. B. (1943). The description of personality: Basic traits resolved into clusters.
Journal of Abnormal Social Psychology. 38.476-506.
Cattell, R. B. (1945). The description of personality: Principles and findings in a factor
analysis. American Journal of Psychology. 58. 69-90.
Cattell, R. B. (1946). The Description and Measurement of Personality. Yonkers, NY:
Worid Book.
Cattell. R. B. (1947). Confirmation and clarification of primary personality factors.
Psvchometrika. 12. 197-220.
Cattell. R. B. (1948). The primary personality factors in women compared with those in
men. British Journal of Psychology. 1. 114-130.
Cattell, R. B. (1965). The Scientific Analysis of Personality. London: Penguin.
Cattel. R. B.. Eber, H. & Tatsuoka, M. (1970). Handbook for the 16PF Questionnaire.
Champaign: IP AT.
Clark. L. A. & Watson. D. (1991). General affective dispositions in physical and
psychological health. In C. R. Snyder & D. R. Forsyth (Eds.), Handbook of Social
and Clinical Psychology (pp. 221-245). New York: Pergamon Press.
Corey. S. M. (1937). Professed attitudes and actual behavior. Journal of Educational
Psychology. 28. 271-280.
Costa, P. T., Jr. & McCrae, R. R. (1992). Revised NEQ Personality Inventory and NEO
Five-Factor Inventory Professional Manual. Odessa, FL: PAR.
Costa. P. T., Jr., McCrae, R. R., & Dye, D. A. (1991). Facet scales for Agreeableness and
Conscientiousness: A revision of the NEO Personality Inventory. Personality and
Individual Differences, 12. 887-898.
137
Cranny, C. J.. Smith, P. C., & Stone, E. F. (1992). Job Satisfaction: How People Feel
About Their Jobs and How It Affects Their Performance. New York: Lexington
Books.
Cudeck. R. (1989). The analysis of correlation matrices using covariance structure models.
Psychological Bulletin. 105. 317-327.
Davidson, A. & Jaccard, J. (1976). Determinants of fertility intentions. In S. Nevmian & V.
Thompson (Eds.), Population psychology: Research and Eduacational Issues.
Washingtion, DC; U.S. Government Printing OfBce.
Davidson, A. & Jaccard, J. (1979). Variables that moderate the attitude-behavior relation;
Results of a longitudinal survey. Journal of Personality and Social Psvchologv.
37(8), 1364-1376.
Decker. P. J. & Borgen, F. H. (1993). Dimensions of work appraisal: Stress, strain, coping,
job satisfaction, and negative affectivity. Journal of Counseling Psychology. 40(4).
470-478.
Digman, J. M. (1990). Personality structure: Emergence of the five-factor model. Annual
Review of Psychology. 41.417-440.
Dipboye, R. L.. Smith, C. S., & Howell. W. C. (1994). Understanding Industrial and
Organizational PsvchQlog\" An Integrated Approach. Fort Worth, TX: Harcourt
Brace.
Doob. L. W. (1947). The behavior of attitudes. Psychological Bulletin. 54. 135-156.
Eysenck, H. J. (1970). The Structure of Human Personality (3rd ed.). London: Methuen.
Eysenck, H. J. & Eysenck, S. J. (1964). The Manual for the Evsenck Personality Invento^\^
San Diego: Educational and Industrial Testing Service.
Ferris. G. R.. Youngblood, S. A., & Yates, V. L. (1985). Personality, training performance,
and withdrawal; A test of the person-group fit hypothesis for organizational
newcomers. Journal of Vocational Behavior. 27. 377-388.
138
Fishbein, M. & Ajzen, I. (1974). Attitudes toward objects as predictors of single and
multiple behavioral criteria. Psychological Review. 81. 59-74.
Fishbein. M. & Ajzen, I. (1975). Belief Attitude. Intention, and Behavior: An Introduction
to Theory and Research. Reading, MA: Addison-Wesley.
Fishbein. M. & Ajzen, I. (1981). Attitudes and voting behavior: An application of the
theory of reasoned action. In G. M. Stephenson & J. M. Davis (Eds.). Progress in
Applied Social Psychology CVol. 1, pp. 253-313). London: Wiley.
Fisher, C. D. & Locke, E. A. (1992). The new look in job satisfaction research and theory.
In C. J. Cranny, P. C. Smith, and E. F. Stone (Eds.), Job Satisfaction: How People
Feel About Their Jobs and How It Affects Their Performance. New York; Lexington
Books.
Fiske, D. W. (1949). Consistency of the factorial structures of personality ratings from
differenct sources. Journal of Abnormal Social Psychology. 44. 329-344.
Fumham, A. (1992). Personality at Work: The Role of Individual Differences in the
Workplace. New York: Routledge.
Fumham, A. & Zacherl, M. (1986). Personality and job satisfaction. Personality and
Individual Differences, 7, 453-459.
George. J. M. (1989). Mood and absence. Journal of Applied Psychology. 74. 317-324.
George. J. M. (1990). Personality, affect and behavior in groups. Journal of Applied
Psychology. 75. 107-116.
George, J. M. (1991). State or trait: Effects of positive mood on prosocial behaviors at work.
Journal of Applied Psychology. 76.299-307.
George, J. M. (1992). The role of personality in organizational life: Issues and evidence.
Journal of Management. 18(2). 185-213.
139
George, J. M. & Bettenhausen. K. (1990). Understanding prosocial behavior, sales
performance, and turnover; A group level analysis in a service context. Journal of
Applied Psvchologv. 75, 698-709.
Gergen, K J., Gergen. M. M., & Meter. K.. (1972). Individual orientations to prosocial
behavior. Journal of Social Issues. 28(3). 105-130.
Gerhart, B. (1987). How important are dispositional factors as determinants of job
satisfaction? Implication for job design and other persoimel programs. Journal of
Applied Psvchologv. 72(3). 366-373.
Getman, J. G., Goldberg, S. B., & Herman, J. B. (1976). Union Representation Elections:
Law and Reality. New York; Russell Sage.
Ghiselli, E. & Barthol, R. (1953). The validity of personality inventions in the selection of
employees. Journal of Applied Psvchologv. 37, 18-20.
Guilford. J. P. (1975). Factors and factors of personality. Psvchological Bulletin. 82. 802-
814.
Guion. R. M. (1991). Personnel assessment, selection, and placement.. In M. D. Duimette
& L. A. Hough (Eds.), Handbook of Industrial and Organizational Psvchologv. Vol
II, (pp.325-397).
Guion, R. M. (1991). Agenda for research and action. In C. J. Crarmy, P. C. Smith, and E.
F. Stone (Eds.). Job Satisfaction; How People Feel About Their Jobs and How It
Affects Their Performance. New York; Lexington Books.
Guion. R. M. & Gottier. R. F. (1965). Validity of personality measures in personnel
selection. Personnel Psvchologv. 18. 135-164.
Hackett, R. D. (1989). Work attitudes and employee absenteeism; A synthesis of the
literature. Journal of Occupational Psvchologv. 62(3), 235-248.
Hackman, J. R. & Oldham, G. R. (1976). Motivation through the design of work; A test of
a theorv'. Organizational Behavior and Human Performance, 16, 250-279.
140
Hakel. M. D. (1974). Normative personality factors recovered from ratings of personality-
descriptors: The beholder's eye. Personnel Psvchology. 27. 409-421.
Hanisch, K. A. (1995a). Behavioral families and multiple causes: Matching the complexity
of responses to the complexity of antecedents. Current Directions in Psychological
Science. 5, 156-162.
Hanisch, K. A.. (1995b). Organizational withdrawal. In N. Nicholson, R. Schuler, & A. Van
de Ven (Eds.), The Dictionary of Organizational Behavior. Oxford, England:
Blackwell Publishers. Hanisch, K. A. & Hulin, C. L. (1990). Job attitudes and organizational withdrawal: An
examination of retirement and other voluntary withdrawal behaviors. Journal of
Vocational Behavior. 37. 60-78.
Hanisch, K. A. & Hulin, C. L. (1991). General attitudes and organizational withdrawal; An
evaluations of a causal model. Journal of Vocational Behavior. 39. 110-128.
Hanisch, K.A., Hulin, C.L., & Roznowski, M. (In Press). Evaluating individuals' repertoires
of behaviors: Scientific reasons and evidence for studying multiple behaviors.
Journal of Organizational Behavior.
Hathaway. S. R. & McKinley, J. C. (1943). Manual for the Minnesota Multiphasic
Personality Inventory. New York: Psychological Corporation.
Hogan, R. T. (1991). Personality and personality measurement. In M. D. Dimnette & L. A.
Hough (Eds.), Handbook of Industrial and Organizational Psychology. Vol II.
(pp.873-919).
Hough, L. M.. Eaton, N. K., Durmette, M. D., Kamp, J. D., & McCIoy. R. A. (1990).
Criterion-related validities of personality constructs and the effect of response
distortion on diose validities. Journal of Applied Psvchology. 75(5), 581 -595.
141
Hulin, C. L. (1991). Adaptation, persistence, and commitment in organizations. In M. D.
Dunnette & L. A. Hough (Eds.), Handbook of Industrial and Organizational
Psychology. Vol II, (pp.445-505).
Hulin, C. L. & Mayer, L. J. (1986). Psychometric equiyalence of a translation of the Job
Descriptiye Index into Hebrew. Journal of Applied Psychology. 71(1) pp. 83-94.
Hulin C. L. & Rousseau, D. M. (1980). Analyzing infrequent events: Once you find them
your troubles begin. In K. H. Roberts & Burstein (Eds.), Issues in aggregation: New
directions for methodology of social and behavioral science. 6. San Francisco:
Jossey-Bass
Huliru C. L., Roznowski, M. & Hachiya, D. (1985). Alternative opportunities and
withdrawal decisions: Empirical and theoretical discrepancies and an integration.
Psychological Bulletin. 97. 233-250.
Hunter, J. E. (1986). Cognitive ability, cognitive aptitudes, job knowledge, and job
performance. Journal of Vocational Behavior. 29. 340-362.
laffaldano, M. T. & Muchinsky, P. M. (1985). Job satisfaction and job performance: A
meta-analysis. Psychological Bulletin. 97.251-273.
Jackson. D. N. (1984). Personality Research Form Manual. (3rd ed.). Port Huron. MI:
Research Psychologists Press.
Jenkins. J. M. (1993). Self-monitoring and turnover: The impact of personality on intent
leave. Journal of Organizational Behavior. 14( 1), 83-91.
Joreskog, K. G. & Sorbom, D. (1989). LISREL 7: A Guide to the Program and
Applications. Chicago: Statistical Package for the Social Sciences.
Joreskog, K. G. & Sorbom, D. (1993). LISREL 8 User's Reference Guide. Chicago:
Scientific Software Intemational, Inc.
142
Judge. T. A. (1993). Does affective disposition moderate the relationship between job
satisfaction and voluntary turnover? Journal of Applied Psychology. 78(3) pp. 395-
401.
Judge. T. A. & Hulin, C. L. (1993). Job satisfaction as reflection of disposition: A Multiple
source causal analysis. Organizational Behavior and Human Decision Processes. 56.
388-421.
Katzell, R. A. (1994). Contemporary meta-trends in industrial and organizational
psychology. In M. D. Dunnette & L. A. Hough (Eds.), Handbook of Industrial and
Organizational Psvcholoev. Vol IV. (pp. 1-94).
Katzell, R, A., Thompson, D. E., & Guzzo, R. A. (1992). How job satisfaction and job
performance are and are not linked. In C. J. Cranny, P. C. Smith, and E. F. Stone
(Eds.), Job Satisfaction: How People Feel About Their Jobs and How It Affects Their
Performance. New York: Lexington Books.
Kemery. E. R. (1991). Affective disposition, role stress, and job withdrawal. Journal of
Social Behavior and Personality. 6(7), 331-347.
Klages. L. (1926). The Science of Character. (Trasl. 1932). London: Allen &Unwin.
Krebs. D. L. (1970). Altruism: An examination of the concept and review of the literature.
Psychological Bulletin, 73,258-302.
Kunin. T. (1955). The construction of a new type of attitude measure. Personnel
Psychology. ^ 65-78.
LaPiere. R. (1934). Attitudes versus actions. Social Forces. 13. 230-237.
Levin I. & Stokes. J. P. (1989). Dispositional approach to job satisfaction: Role of negative
affectivity. Joumal of Applied Psychology. 74(5) pp. 752-758.
Locke. E. A. (1976). The nature and causes of job satisfaction. In M. D. Dunnette (Ed.),
Handbook of Industrial and Organizational Psychology (pp. 1297-13431. Chicago:
Rand McNally.
143
Lubinski, D. & Thompson, T. (1986). Functional units of human behavior: A dispositional
analysis. In T. Thompson & M. Zeiler (Eds.), Analvsis and Integration of Behavioral
Units (pp. 315-334). Hillsdale. NJ: Erlbaum.
March, J. G. & Simon, H. A. (1958). Organizations. New York: Wiley.
McClelland. D.. Atkinson, J.. Clark. R.. & Lowell, E. (1953). The Achievement Motive.
New York: Appleton-Century-Crofts.
McHenry, J. J., Hough, L. M., Toquam, J. L., Hanson. M. A.. & Aschworth, S. A. (1990).
Project A validation results: The relationship between predictor and criterion
domains. Personnel Psvchologv. 43. 335-353.
Mobley, W. H., Horner, S. O., & Hollingsworth, A. T. (1978). An evaluation of precursors
of hospital employee turnover. Journal of Applied Psvchologv. 63.408-414.
Motowidlo, S. J., (1984). Does job satisfaction lead to consideration and personal
sensitivity? Academy of Management Journal. 27. 910-915.
Mowday, R. T. & Spencer, D. G. (1981). The influence of task and personality
characteristics on employee turnover and absenteeism incidents. Academv of
Management Journal. 24(3), 634-642.
Murray, H. A. (1938). Explorations in Personality. New York: Oxford University Press.
Newton. T. & Keenan. T. (1991). Further analyses of the dispositional argument in
organizational behavior. Journal of Applied Psvchologv. 76(6), 781-787.
Norman. W. T. (1963). Toward an adequate taxonomy of personality attributes: Replicated
factor structure in peer nomination personality ratings. Abnormal Social Psychology.
6^ 574-583.
Organ, D. W. (1988). Organizational Citizenship Behavor: The Good Soldier Syndrom.
Lexington. MA: Lexington Books.
Organ, D. W. (1994). Personality and organizational citizenship behavior. Journal of
Management. 20(2). 465-478.
144
Organ. D. W. & Konovsky, M. A. (1989). Cognitive versus affective determinants of
organizational citizenship behavior. Journal of Applied Psvchologv. 74(1). 157-164.
Organ. D. W. & Lingl, A. (1995). Personality, satisfaction, and organizational citizenship
behaivor. The Journal of Social Psvchologv. 135(3). 339-350.
Orpen, C. (1971). The fakability of the Edwards Personal Preference Schedule in personnel
selection. Persoimel Psvchoogv. 24. 1-4.
Parsons. C. K. & Hulin, C. L. (1982). An empirical comparison of item response theory and
hierarchical factor analysis in applications to the measurement of job satisfaction.
Journal of Applied Psvchologv. 67. 826-834.
Pedhazur, E. J. & Schmelkin, L. P. (1991). Measurement. Design, and Analvsis: An
Integrated Approach. Hillsdale, NJ: Lawrence Erlbaum.
Petty, R. E. & Cocioppo, J. T. (1981). Attitudes and Persuasion: Classic and Contemporarv
Approaches. Dubuque, lA: William C. Brown.
Podsakoff, P. M. & Williams, L. (1986). The relationship between job pertbrmance and job
satisfaction. In E. A. Locke (Ed.). Generalizing From Laboratory to Field Settings
(pp. 207-253). Lexington, MA; Lexington Books.
Podsakoff. P. M.. MacKenzie, S. B.. Moorman, R. H.. & Fetter, R. (1990). Transformational
leader behaviors and their effects on followers' trust in leader, satisfaction, and
organizational citizenship behavior. Leadership Ouarterlv. 1, 107-142.
Puffer, S. M. (1987). Prosocial behavior, non-compliant behavior, and work performance
among commission salespeople. Joumal of Applied Psvchologv. 72. 615-621.
Price. J. L. (1977). The Study of Turnover. Ames, Iowa: Iowa State University Press.
Rosse, J. G. & Miller, H. E. (1984). Relationship between absenteeism and other employee
behaviors. In P. S. Goodman & R.S. Atkin (Eds.), Absenteeism: New Approaches to
Understanding. Measuring, and Managing Employee Absence. San Francisco:
Jossey-Bass.
145
Rosse, J. G. Miller, H. E., & Barnes, L. K. (1991). Combining personality and cognitive
ability predictors for hiring service-oriented employees. Journal of Business and
Psvchologv. 5(4), 431-445.
Roznowski, M. (1989). Examination of the measurement properties of the job descriptive
index with experimental items. Journal of Applied Psychology. 74(5). 805-814.
Roznowski, M. & Hanisch, K. A. (1990). Building systematic heterogeneity into attitude and
behavior measures. Journal of Vocational Behavior. 36. 361-375.
Roznowski, M. & Hulin, C. L. (1992). The scientific merit of valid measures of general
constructs with special reference to job satisfaction and job withdrawal. In C. J.
Cranny, P. C. Smith, and E. F. Stone (Eds.), Job Satisfaction: How People Feel
About Their Jobs and How It Affects Their Performance. New York: Lexington
Books.
Roznowski, M., Rosse, J. G., & Miller, H. E. (1992). The Utility of Broad-Band Measures of
Employee Behavior: The Case for Employee Adaptation and Citizenship. Paper
presented at the 52nd annual meeting of the Academy of Management. Las Vegas.
Rushton, J. P., Brainerd, C. J., & Pressley, M. (1983). Behavioral development and construct
validity: The principle of aggregation. Psychological Bulletin. 94(1). 18-38.
Salancik, G. R. & Pfeffer, J.(1977). An examination of need-satisfaction models of job
attitudes. Administrative Science Quarterly. 22,427-456.
Salancik, G. R. & Pfeffer, J. (1978). A social information processing approach to job
attitudes and task design. Administrative Science Quarterly. 23. 224-251.
Schmidt, F. L. & Hunter, J. E. (1981). Employment testing: Qld theories and new research
findings. American Psychologist, 36. 1128-1137.
Schneider, B. & Dachler, H. P. (1978). A note on the stability of the job description index.
Journal of Applied Psychology. 63. 650-653.
146
Schneider, R. J. & Hough, L. M. (1995). Personality and industrial/organizational
psychology. In C. L. cooper and L. T. Robertson (Eds.), International Review of
Industrial and Organizational Psychology. (Vol. 10). New York: John Wiley &
Sons Ltd.
Schriesheim. C. (1978). Job satisfaction, attitudes toward unions, and voting in a union
representation election. Journal of Applied Psychology. 63. 548-552.
Schumacker, R. E. & Lomax, R. G. (1996). A Beginner's Guide to Structural Equation
Modeling. Mahwah, NJ: Lawrence Erlbaum.
Scott, K. D. & Taylor, G. S. (1985). An examination of conflicting findings on the
relationship between job satisfaction and absenteeism: A meta-analysis. Academy of
Management Journal. 28. 599-612.
Smith. C. A., Organ, D. W., & Near, J. P. (1983). Organizational citizenship behavior: Its
nature and antecedents. Journal of Applied Psychology. 68(4). 653-663.
Smith, G. M. (1967). Usefulness of peer ratings of personality in educational research.
Educational Psychology Measurement. 27, 967-984.
Smith, P. C. (1992). In pursuit of happiness: Why study general job satisfaction? In C. J.
Cranny, P. C. Smith, and E. F. Stone (Eds.), Job Satisfaction: How People Feel
About Their Jobs and How It Affects Their Performance. New York: Lexington
Books
Smith, P.C.. Balzer, W., Brannick, M., Eggleston, S., Gibson, W., Ironson, G., Josephson, H..
Paul, K., Reilly, C., & Whalen, M. (1987). Manual for the revised JDI and Job in
General scales. Bowling Green, OH: Department of Psychology, Bowling Green
State University.
Smith, P. C.. Kendall, L., & Hulin C. L. (1969). The Measurement of Satisfaction in Work
and Retirement. Chicago: Rand McNally.
147
Staw. B. M. & Ross, J. (1985). Stability in the midst of change: A dispositional approach to
job attitudes. Joumal of Applied Psychology. 70(3). 391-407.
Staw. B. M., Bell, N. E., & Clausen, J. A. (1986). The dispositional approach to job
attitudes: A lifetime longitudinal test. Administrative Science Quarterly. 31. 56-77.
Steers, R. M. & Rhodes, S. R. (1978). Major influences on employee attendance: A process
model. Joumal of Applied Psychology. 63. 391-407.
Tellegen, A. (1985). Stmctures of mood and personality and their relevance to assessing
anxiety, with an emphasis on self-report.
Tett, R. P. & Meyer, J. P. (1993). Job satisfaction, organizational commitment, tumover
intention, and tumover: Path analyses based on meta-analytic findings. Personnel
Psychology. 46.259-293.
Tett, R. P., Jackson, D. N., & Rothstein, M. (1991). Personality measures as predictors of job
performance: A meta-analytic review. Personnel Psychology. 44. 703-742.
Thibaut, J. W. & Kelley, H. H. (1959). The Social Psychology of Groups. New York:
Wiley.
Thurstone, L. L. (1931). The measurement of social attitudes. Joumal of Abnormal and
Social Psychology. 26. 249-269.
Triandis, H. C. (1971). Attitudes and Attitude Change. New York: Wiley.
Tupes. E. C. (1957). Personality traits related to effectiveness of junior and senior Air Force
officers (USAF Personnel Training Resource Center Resource Rep. No. 57-125)
Tupes, E. C. & Christal, R. E. (1961). Recurrent personality factors based on trait ratings
(USAF Technical Rep. No. 61-97).
Underwood, B. & Moore, B. (1982). Perspective-taking and altruism. Psychological
Bulletin. 91. 143-173.
Vroom. V. (1964). Work and Motivation. New York: Wiley.
148
Waller, N. G. & Ben-Porath, Y. S. (1987). Is it time for clinical psychology to embrace the
Five-Factor Model of personality? American Psvcholoeist. 42, 887-889.
Watson, D. & Clark, L. A. (1984). Cross-cultural convergence in the structure of mood: A
Japanese replication and a comparison with U. S. findings. Journal of Personality and
Social Psychology. 47. 127-144
Watson, D., Clark, L. A. & Tellegen, A. (1988). Development and validation of brief
measures of positive and negative affect: The Panas scales. Journal of Personality
and Social Psychology. 54. 1063-1070.
Weigel, R. H., Vernon, D. T., & Tognacci, L. N. (1974). Specificity of the attitude as a
determinant of attitude-behavior congruence. Journal of Personality and Social
Psychology. 30. 724-728.
Weiss, D. J. & Adler, S. (1984). Personality and organizational behavior. In B. M. Staw &
L. L. Cummings (Eds.). Research in Organizational Behavior. Vol. 6, (pp. 1-50).
New York: JAI Press.
Weiss. D. J., Dawis, R. B., England, G. W., & Lofqiiist, L. H. (1967). Manual for the
Minnesota Satisfaction Questionnaire. Minneapolis: Industrial Relations Center,
University of Minnesota.
Wicker, A. W. (1969). Attitudes versus actions: The relationship of verbal and overt
behavioral responses to attitude objects. Journal of Social Issues. 25.41-78.
Wicker, A. W. (1971). An examination of the "other-variables" explanation of attitude-
behavior inconsistency. Journal of Personality and Social Psychology. 19. 18-30.
IMAGE EVALUATION TEST TARGET (QA-3)
1.0
I.I
1.25
m Ih | |2 .2
1 ̂11 ̂
1 ̂li£
II i— II
1.4 1.6
1 SOtTirn
6 w
V
<p 7
> //
o 7
•/. >qPPUED A IIVA4GE . Inc
—— 1653 East Main Street -=^- Rochester. NY 14609 USA
Phone: 716/482-0300 Fax: 716/288-5989
01993. Anplled Image. Inc.. Ail Rights Reserved